Level I of the CFA® 2025 Exam

Study Notes - Quantitative Methods

Offered by AnalystPrep

Last Updated: Sep 15, 2024

1
©2024 AnalystPrep “This document is protected by International copyright laws. Reproduction and/or distribution of this document is

prohibited. Infringers will be prosecuted in their local jurisdictions.”

 Table of Contents

1 - Rate and Return 3

2 - The Time Value of Money in Finance 31
3 - Statistical Measures of Asset Returns 74
4 - Probability Trees and Conditional Expectations 109
5 - Portfolio Mathematics 126
6 - Simulation Methods 145
7 - Estimation and Inference 162
8 - Hypothesis Testing 178
9 - Parametric and Non Parametric Tests of Independence 217
10 - Simple Linear Regression 230
11 - Introduction to Big Data Techniques 281

2
© 2014-2024 AnalystPrep.
 Learning Module 1: Rate and Return

LOS 1a: interpret interest rates as required rates of return, discount

rates, or opportunity costs and explain an interest rate as the sum of a
real risk-free rate and premiums that compensate investors for bearing
distinct types of risk

The time value of money is a concept that states that cash received today is more valuable than

cash received in the future. If a person agrees to receive payment in the future, he foregoes the

option of earning interest if he invests that amount of money today.

An interest rate or yield, usually denoted by r, is a rate of return that reflects the connection

between cash flows dated at different times.

Assume you currently possess $100. Next, consider depositing this money into a savings account,

expecting it to grow to $110 after one year. Intuitively, the compensation required for deferring

the consumption of $100 now in favor of receiving $110 in one year is $10 (equal to 110 minus

100). This compensation is equivalent to a 10% rate of return (calculated as 10 divided by 100).

There are three ways to interpret interest rates:

1. Required rate of return: The minimum return an investor expects to earn to accept an

investment.

2. Discount rate: The rate used to discount future cash flows to allow for the time value of

money (to determine the present value equivalent of some money to be received

sometime in the future). Discount rates and interest rates are used almost

interchangeably.

3. Opportunity cost: The value of the best-forgone alternative; the most valuable

alternative investors give up when they choose what to do with money.

Determinants of Interest Rates

Economics postulates that the forces of supply and demand determine interest rates. In this

case, the investors (lenders) supply the money, and the borrowers demand money for their

3
© 2014-2024 AnalystPrep.
consumption.

As such, interest is a borrower's reward for using an asset, usually capital, belonging to a lender.

It is compensation for the loss or value depreciation occasioned by the use of the asset.

Therefore, an interest rate is composed of a real risk-free interest rate plus a set of four

premiums that represent compensation for bearing distinct types of risk:

Real risk-free interest rate

+ Inflation Premium
Interest (r) = + Default risk premium
+ Liquidity premium
+ Maturity premium

The Real Risk-free Interest Rate

The real risk-free interest rate is the single-period interest rate for a completely risk-free security

if no inflation is expected. According to economic theory, the real risk-free rate reflects people’s

preferences for current compared to real future consumption.

Types of Risk Premiums

Inflation Risk Premium

Inflation risk is the loss of money's purchasing power caused by increased consumer goods

prices.

The inflation premium compensates investors for expected inflation. It represents the average

inflation rate expected over the debt's maturity. The risk of a decrease in purchasing power

validates the inflation risk premium.

Liquidity Risk Premium

Liquidity refers to the ease with which an investment can be converted into cash without

significantly sacrificing market value.

4
© 2014-2024 AnalystPrep.
The liquidity premium compensates investors for the risk of loss relative to an investment’s fair

value if the investment needs to be converted to cash quickly.

Default Risk Premium

Default risk describes a situation in which a borrower may fail to repay borrowed funds due to

bankruptcy, which might result in significant losses on the lender's part.

The default risk premium compensates investors for the possibility that the borrower will fail

to make a promised payment at the contracted time and in the contracted amount.

Maturity Risk Premium

The maturity risk premium is the additional return an investor requires for assuming interest

rate and reinvestment risk resulting from a longer investment maturity timeline. Maturity risk

premium increases with an increase in the maturity timeline. In other words, the longer the

maturity timeline of an investment, the higher the maturity risk premium.

Nominal Risk-free Interest Rate

The nominal risk-free interest rate is defined as the sum of the real risk-free interest rate and the

inflation premium. In other words, the nominal risk-free interest rate can be seen as the

combination of the real risk-free rate plus an inflation premium, as shown by the following

equation:

(1+Nominal risk-free rate)=(1+Real risk-free rate )(1+Inflation premium)

The above equation is generally approximated as follows:

Nominal risk-free rate=Real risk-free rate+Inflation premium

Most rates quoted on short-term government debts can be taken as nominal risk-free interest

rates over the respective maturity.

5
© 2014-2024 AnalystPrep.
Question

Which of the following is most likely an interpretation of interest rate as a benefit

foregone when investors spend money on current consumption instead of saving or

investing?

A. Discount rate.

B. Opportunity cost.

C. Required rate of return.

Solution

The correct answer is B.

Opportunity cost is a key factor in interpreting interest rates. It refers to the interest

foregone when investors opt for an alternate option, such as spending on current

consumption instead of saving or investing.

A is incorrect. The discount rate is the interest rate used to discount future cash

flows to reach the present value.

C is incorrect. The required rate of return is the minimum rate of return an investor

would wish to earn to postpone current consumption.

6
© 2014-2024 AnalystPrep.
LOS 1b: calculate and interpret different approaches to return
measurement over time and describe their appropriate uses

Financial assets are primarily defined based on their return-risk characteristics. This definition

approach helps when building a portfolio from all the assets available. It's noteworthy that there

are different ways of measuring returns.

Financial market assets generate two types of returns: Income from cash dividends or interest

payments and capital gains or losses from changes in the prices of financial assets.

Some financial assets give only one stream of return. For instance, headline stock market indices

typically only report on price appreciation. They do not include the dividend income unless the

index clarifies that it is a “total return” series.

Holding Period Return

A holding period return is earned from holding an asset for a specified period, such as a day,

month, or ten years.

The general formula of the holding period return is given by:

(P 1 − P0 ) + I1
R=
P0

P 0 = Price of an asset at the beginning of the period (t=0).

P 1 = Price of an asset at the end of the period (t=1).

I1 = Income received at the end of the period (t=1).

Example: Calculating Holding Period Return

An investor purchased 100 shares of a stock at $50 per share and held the investment for one

year. During that period, the stock paid dividends of $2 per share. At the end of the year, the

investor sold all the shares for $60 per share.

7
© 2014-2024 AnalystPrep.
The holding period return is closest to:

Solution

In this case, we have:

P 0 = 100 shares × $50 per share = $5,000

I1 = 100 shares × $2 per share = $200
P 1 = 100 shares × $60 per share = $6,000

Therefore,

(P1 − P 0 ) + I1 6,000 − 5,000 + 200

R= = = 24%
P0 5, 000

Holding period returns can also be calculated for periods longer than a year. For instance, if we

need to calculate the holding period return for a five-year period, we should compound the five

annual returns as follows:

(P5 − P 0 ) + I(1−5)
R=
P0

Arithmetic Return

When we have assets with multiple holding periods, we must aggregate the returns into one

overall return.

Denoted by R̄ i arithmetic mean for an asset i is a simple process of finding the average holding

period returns. It is given by:

Ri, 1 + R i,2 + … + Ri ,T−1 + R iT 1 T

R̄ i = = ∑ Ri
T T t=1 t

Where:
R it = Return of asset i in period t.

8
© 2014-2024 AnalystPrep.
T = Total number of periods.

For example, if a share has returned 15%, 10%, 12%,

and 3% over the last four years, then the arithmetic
mean is computed as follows:

1 T 1
R̄ i = ∑ R it = (15% + 10% + 12% + 3%) = 10%
T t=1 4

Geometric Return

Computing a geometric mean follows a principle similar to the one used to compute compound

interest. It involves compounding returns from the previous year to the initial investment's value

at the start of the new period, allowing you to earn returns on your returns.

A geometric return provides a more accurate representation of the portfolio value growth than

an arithmetic return.

Denoted by R̄ Gi the geometric return for asset i is given by:

R̄G i = √T (1 + Ri ,1 ) × (1 + Ri2 ) × … × (1 + Ri ,T−1 ) × (1 + Ri T ) − 1

T
= ⎷T ∏ (1 + R t) − 1
t=1

Using the same annual returns of 15%, 10%, 12%, and 3% as shown above, we compute the

geometric mean as follows:

1
Geometric mean = [(1 + 15%) × (1 + 10%) × (1 + 12%) × (1 + 3%)] 4 − 1
= 9.9%

Note that the geometric return is slightly less than the arithmetic return. Arithmetic returns tend

to be biased upwards unless the holding period returns are all equal.

Harmonic Mean

9
© 2014-2024 AnalystPrep.
The harmonic mean measures central tendency. It's especially useful for rates or ratios such as

P/E ratios. The harmonic mean's formula is derived from the harmonic series, a specific

mathematical sequence.

n
X̄ H = , Xi > 0 for all i = 1, 2, … , n
∑ni=1 X1
i

The above formula is interpreted as the “harmonic mean of observations X1 , X2 , … , Xn .”

The harmonic mean is handy for averaging ratios when those ratios are consistently applied to a

fixed quantity, resulting in varying unit numbers. For instance, it's applied in cost-averaging

strategies where you invest a fixed amount of money at regular intervals.

Example: Calculating the Harmonic Mean

An investor practices cost averaging by investing in a particular stock over a three-month period.

The investor decides to allocate different amounts of money each month. In the first month, the

investor invests $2,000; in the second month, $3,000; and in the third month, $4,000. The share

prices of the stock for these three months are $10, $12, and $15, respectively.

Calculate the average price paid per share for the three-month period.

Solution

Using the harmonic mean formula,

n 3
X̄ H = = = 12
∑ ni=1 X1 1 1 1
+ +
i 10 12 15

Trimmed and Winsorized Means

Trimmed and Winsorized means seek to lower the effect of outliers in a data set.

Trimmed Mean

10
© 2014-2024 AnalystPrep.
The trimmed mean is a measure of central tendency. We calculate it after excluding a small

percentage of the lowest and highest values from a dataset.

For example, a data set consists of 10 observations: 12, 15, 18, 20, 22, 25, 27, 30, 35, and 40. We

can calculate the trimmed mean after removing the highest and lowest values.

After removing these values, the remaining data set is 15, 18, 20, 22, 25, 27, 30, and 35.

Now, let’s calculate the trimmed mean by taking the average of these remaining values:

15 + 18 + 20 + 22 + 25 + 27 + 30 + 35192
= = 24
8 8

Therefore, the trimmed mean of the given data set is

24.

Winsorized Mean

The Winsorized mean is a central tendency measure. It works by replacing extreme values at

both ends of the data with the values of their closest observations. This process is similar to the

trimmed mean. Essentially, it helps eliminate outliers in a dataset.

For example, consider a dataset of 12 observations: 8,

12, 15, 18, 20, 22, 25, 27, 30, 35, 40, and 50. We can
calculate the Winsorized mean by replacing the
lowest and highest values with those closest to the
10th and 90th percentiles, respectively. As such, the
new values are 10, 12, 15, 18, 20, 22, 25, 27, 30, 35,
37.5, and 40, and the winsorized mean is:
10 + 12 + 15 + 18 + 20 + 22 + 25 + 27 + 30 + 35 + 37.5 + 40
≈ 24.46
12

11
© 2014-2024 AnalystPrep.
Question 2

What are the arithmetic mean and geometric mean, respectively, of an investment

that returns 8%, -2%, and 6% each year for three years?

A. Arithmetic mean = 5.3%; Geometric mean = 5.2%.

B. Arithmetic mean = 4.0%; Geometric mean = 3.6%.

C. Arithmentic mean = 4.0%; Geometric mean = 3.9%.

Solution

The correct answer is C.

8% + (−2%) + 6%
Arithmetic mean = = 4%
3

Geometric mean = [(1 + 8%) × (1 + (−2%)) × (1 + 6%)]1/3 − 1 = 3.9%

12
© 2014-2024 AnalystPrep.
LOS 1c: Compare the money-weighted and time-weighted rates of return
and evaluate the performance of portfolios based on these measures

Money-weighted Rate of Return

The money-weighted return considers the money invested and gives the investor information on

the actual investment return. Calculating money-weighted return is similar to calculating an

investment’s internal rate of return (IRR).

The money-weighted rate of return (MWRR) is like the portfolio's internal rate of return (IRR).

It's the rate at which the present value of cash flows equals zero. It's a way to measure how well

a portfolio performs.

T
CFt
∑ =0
t=0 (1 + I RR)t

Where:

T = Number of periods.

CFt = Cash flow at time t.

I RR = Internal rate of return (or money-weighted rate of return).

The money-weighted rate of return (MWRR) looks at a fund’s starting and ending values and all

the cash flows in between. In an investment portfolio, cash inflows are a part of it. These inflows

could be from deposits or investments made during a certain period. The MWRR considers these

inflows and calculates the overall rate of return for the portfolio:

The beginning value.

Dividends/interest reinvested.

Contributions made.

Cash outflows, on the other hand, refer to:

13
© 2014-2024 AnalystPrep.
 Withdrawals made.

Dividends or interest received.

The final value of the fund.

Example 1: Calculating Money-weighted Rate of Return

An investor makes the following investments in a portfolio over a two-year period:

At the beginning of year one, the investor invests $10,000.

At the end of the first year, after the portfolio's value increases to $12,000, the investor

adds $5,000, making the total portfolio value $17,000.

At the end of the second year, the portfolio value further increases to $25,000.

The money-weighted rate of return for the investor’s portfolio is closest to:

Solution

We need to calculate the internal rate of return (IRR) considering the following cash flows:

CF0 = −$10, 000 (Initial investment).

CF1 = −$5, 000 (Additional investment at the end of year one).

CF2 = +$25, 000 (Final portfolio value at the end of year two).

To find the money-weighted rate of return, solve the equation for IRR:

CF0 CF1 CF2 −10, 000 −5, 000 25,000

+ + = + + =0
(1 + IRR)0 (1 + I RR)1 (1 + I RR)2 1 (1 + I RR) (1 + IRR)2

Using BA II Plus Calculator, I RR ≈ 35.08% .

Example: Calculating Money-weighted Return for a Dividend-paying

Stock

14
© 2014-2024 AnalystPrep.
Calvin Hair purchased a share of Superior Car Rental Company for $85 at the beginning of the

first year. He bought an additional unit for $87 at the end of the first year. At the end of the

second year, he sold both shares at $90. During both years, Hair received a dividend of $4 per

share, which was not reinvested.

Calculate the money-weighted return.

Solution

To calculate the money-weighted return in this example, we need to consider the timing and

amounts of cash flows and their respective investment periods.

Step 1: Calculate the total investment at the beginning (t=0):

Initial investment = −$85

Step 2: Calculate the total investment at t = 1:

Initial investment + Additional investment = $87 − $4( Dividend received at

the end of the first year, which is not reinvested)
= −$83

Step 3: Calculate the final portfolio value at t = 2:

Number of shares sold × Selling price = 2 shares × $90 = $180 + 8( Dividend received for
both shares)
= $188

As such, we have:

CF0 = −85 .

CF1 = −83 .

CF2 = 188.

Using the BA II Plus calculator, you will get IRR = 7.71% , equivalent to the money-weighted rate

of return.

15
© 2014-2024 AnalystPrep.
Shortcomings of the Money-weighted Rate of Return

The money-weighted rate of return (MWRR) considers all cash flows, such as withdrawals or

contributions. If an investment spans multiple periods, MWRR gives more importance to the

fund's performance when the account is at its largest. This can be a problem for fund managers

because it might make their performance seem worse due to factors they can't control.

Time-weighted Rate of Return

The time-weighted rate of return (TWRR) calculates an investment's compound growth. Unlike

the money-weighted rate, it doesn't care about withdrawals or contributions. TWRR is like

finding the average return of different time periods within your investment.

Steps of Calculating Time-weighted Rate of Return

Step 1: Value the portfolio immediately before any significant cash inflow or outflow of funds.

Divide the evaluation period into subperiods based on dates of significant additions or

withdrawals of funds.

Step 2: Compute the holding period return on the portfolio for each period.

Step 3: Compound or link the holding period returns to the annual rate of return, which is the

time-weighted rate of return.

TW RR = (1 + HP R 1 × (1 + H P R2 ) × (1 + HP R3 ) … × (1 + HP Rn −1 ) × (1 + H P Rn ) – 1

If the evaluation period is more than one year, compute the geometric mean of the annual

returns to get the time-weighted return for the investment period.

R̄G i = √T (1 + HPR 1 ) × (1 + HPR 1 ) … × (1 + HPRn ) − 1

1
= [(1 + HPR1 ) × (1 + HPR1 ) … × (1 + HPR n )] n − 1

Example: Calculating the Time-Weighted Rate of Return (Period More

than one year)

16
© 2014-2024 AnalystPrep.
An investor purchases a share of stock at t = 0 for $200. At the end of the year (at t = 1), the

investor purchases an additional share of the same stock, this time for $220. She then sells both

shares at the end of the second year for $230 each. She also received annual dividends of $3 per

share at the end of each year. Calculate the annual time-weighted rate of return on her

investment.

Solution

First, we break down the two years into two one-year periods.

Holding period 1:

Beginning value = 200.

Dividends paid = 3.

Ending value = 220.

Holding period 2:

Beginning value = 440 (2 shares × 220)

Dividends paid = 6 (2 shares × 3)

Ending value = 460 (2 shares × 230)

Secondly, we calculate the HPR for each period:

(220 − 200 + 3)
H P R1 = = 11.5%
200
(460 − 440 + 6)
H P R2 = = 5.9%
440

Lastly, we need to find the geometric mean of the HPRs since we are dealing with a period of

more than a year.

1
T WRR = [(1 + HPR 1 ) × (1 + HPR 1 ) … × (1 + HPRn )] n − 1
0.5
= (1.115 × 1.059) − 1 = 8.7%

Example: Calculating the Time-weighted Rate of Return (Period Less

17
© 2014-2024 AnalystPrep.
than One Year)

The beginning value of a portfolio as of January 1, 2020, was $1,000,000. On February 10, the

portfolio’s value was $1,100,000, including an additional contribution of the $50,000 injected

into the portfolio on this date. The portfolio’s ending value at the beginning of April was

$1,350,000.

The time-weighted rate of return is closest to:

Solution

The time-weighted return is calculated as follows:

V1 − V0 (1, 100, 000 − 50 , 000) − 1 , 000, 000

H P R1 = = = 5%
V0 1, 000, 000
V2 − V1 1, 350, 000 − 1, 100, 000
H P R2 = = = 22.73%
V1 1, 100, 000
⇒ T WRR = (1 + HP R 1 ) × (1 + H PR 1 ) − 1
= 1.05 × 1.2273 − 1 = 28.87%

18
© 2014-2024 AnalystPrep.
Question

A chartered analyst buys a share of stock at time t = 0 for $50. At t = 1, he purchases

an extra share of the same stock for $53. The share gives a dividend of $0.50 per

share for the first year and $0.60 per share for the second year. He sells the shares at

the end of the second year for $55 per share. Calculate the annual time-weighted rate

of return.

A. 5.90%.

B.12.24%.

C. 7.00%.

The correct answer is A.

We have two one-year holding periods:

HP 1 HP 2
P 0 = 50 P 0 = 106
D = 0.5 D = 1.2
P 1 = 53 P 1 = 110

We now calculate the holding period returns:

(53 − 50 + 0.5)
HP R 1 = = 7%
50
(110 − 106 + 1.2)
HP R 2 = = 4.9%
106
⇒ T WRR = 1.07 × 1.049 − 1 = 12.24%

Therefore,

Annual TWRR = (1 + 0.1224)0.5 − 1 = 5.9%

19
© 2014-2024 AnalystPrep.
LOS 1d: Calculate and interpret annualized return measures and
continuously compounded returns and describe their appropriate uses

To compare returns over different timeframes, we need to annualize them. This means

converting daily, weekly, monthly, or quarterly returns into annual figures.

Non-Annual Compounding

Interest may be paid semiannually, quarterly, monthly, or even daily – interest payments can be

made more than once a year. Consequently, the present value formula can be expressed as

follows when there are multiple compounding periods in a year:

R S −m N
P V = FVN (1 + )
m

Where:

m = Number of compounding periods in a year.

R s = Quoted annual interest rate.

N = Number of years.

Example: Calculating the Present Value of a Lump Sum (More than One
Compounding Period)

Jane Doe wants to invest money today and have it become $500,000 in five years. The annual

interest rate is 8%, and it's compounded quarterly. How much should Jane invest right now?

Using the formula above:

F V N = $500, 000 .

R S = 8% .

m = 4.

20
© 2014-2024 AnalystPrep.
 8%
R s /m = = 2% = 0.02 .
4

N = 5.

mN = 4 × 5 = 20 .

Therefore,

RS −m N −20
P V = F V N (1 + ) = $500, 000 × (1.02) = $336, 485.67
m

Using BA II Plus Calculator:

Press the [2nd] button, then the [FV] button to clear the financial registers. The display

should show “CLR TVM.”

Enter the future value (FV). This is the amount Jane wants to have in five years, which

is $500,000. To do this, type “500000” and press the [FV] button.

Enter the interest rate (I/Y). This is the annual interest rate, which is 8%. However,

since interest is compounded quarterly, we need to divide this by 4. To do this, type “8”,

press the [÷] button, type “4”, then press the [ENTER] button, and finally press the

[I/Y] button.

Enter the number of periods (N). This is the number of quarters in five years, which is

5*4 = 20. To do this, type “20” and press the [N] button.

Compute the present value (PV). To do this, press the [CPT] and then the [PV] buttons.

The display should show the amount Jane needs to invest today, approximately

$336,485.49.

Annualized Returns

To annualize a return for a period shorter than a year, you need to account for how many times

that period fits into a year. For example, if you have a weekly return, you would compound it 52

times because there are 52 weeks in a year.

21
© 2014-2024 AnalystPrep.
Generally, we can annualize the returns using the following formula:

Returnannual = (1 + Returnperiod )c − 1

Where:

Returnperiod = Quoted return for the period.

c = Number of periods in a year.

Example: Annualizing Returns

If the monthly return is 0.7%, then the compound annual return is:

12
Returnannual = (1 + Returnmonthly) − 1
12
= (1.007) − 1 = 0.0873 = 8.73%

For a period of more than one year, for example, a 15-month return of 16% can be annualized as:

12
Returnannual = (1 + Return 15 month ) 15 − 1
4
= (1.16)5 − 1 = 12.61%

We may apply the same procedure to convert weekly returns to annual returns for comparison

with weekly returns.

52
Returnannual = (1 + Returnweekly ) −1

For comparison with weekly returns, we can convert annual returns to weekly returns by making

(Returnweekly )52 the subject of the formula.

Example: Comparing Investments by Annualizing Returns

An investor is evaluating the returns of two recently formed bonds. Selected return information

on the bonds is presented below:

22
© 2014-2024 AnalystPrep.
 Bond Time Since Issuance Return Since Issuance (%)
A 120 days 2.50
B 8 months 6.00

Annualized Return Calculation

To compare the annualized rate of return for both bonds, you can use the formula for annualizing

returns based on different time periods:

365
Return Since Issuance
Annualized Return = (1 + ) Time Since Issuance
−1
100

Let's calculate the annualized returns for both bonds:

For Bond A:

Time Since Issuance = 120 days.

Return Since Issuance = 2.50%.

365
2.50 120
Annualized Return for Bond A = (1 + ) − 1.
100

Annualized Return for Bond A = (1 + 0.025)3.0417 − 1.

Annualized Return for Bond A = 1.079847 - 1 = 0.079847 or 7.98%.

For Bond B:

Time Since Issuance = 8 months = 240 days.

Return Since Issuance = 6.00%.

365
6.00 240
Annualized Return for Bond B = (1 + ) − 1.
100

Annualized Return for Bond B = (1 + 0.06)1.5208 − 1.

Annualized Return for Bond B = 1.092751 - 1 = 0.092751 or 9.28%.

Comparing the annualized returns:

Bond A has an annualized return of approximately 7.98%.

23
© 2014-2024 AnalystPrep.
Bond B has an annualized return of approximately 9.28%.

Therefore, Bond B has a higher annualized rate of return compared to Bond A.

Continuously Compounded Returns

The continuously compounded return is calculated by taking the natural logarithm of one plus

the holding period return. For example, if the monthly return is 1.2%, you'd calculate it as

ln(1.012), which equals approximately 0.01192.

Generally, continuously compounded from t to t + 1 is given by:

P t+1
rt,t+1 = ln ( ) = ln (1 + Rt,t+1 )
Pt

Assume now that the investment horizon is from time t = 0 to time t = T then the continuously

compounded return is given by:

PT
r0 ,T = ln ( )
P0

If we apply the exponential function on both sides of the equation, we have the following:

PT = P 0 er0,T

PT
Note that can be written as:
P0

PT PT P T−1 P1
=( )( ) …( )
P0 P T −1 P T−2 P0

If we take natural logarithm on both sides of the above equation:

PT PT P T −1 P
ln ( ) = ln ( ) + ln ( ) + … + ln ( 1 )
P0 P T−1 P T −2 P0
⇒ r0 ,T = rT−1, T + rT −2,T −1 + … + r0, 1

24
© 2014-2024 AnalystPrep.
Therefore, the continuously compounded return to time T is equivalent to the sum of one-period

continuously compounded returns.

25
© 2014-2024 AnalystPrep.
Question

The weekly return of an investment that produces an annual compounded return of

23% is closest to:

A. 0.40%.

B. 0.92%.

C. 0.41%.

The correct answer is A.

Recall that:

52
Returnannual = (1 + Returnweekly ) −1

We can rewrite the above equation as follows:

1
Returnweekly = (1 + Return annual) 52 − 1
1
= (1 + 0.23)52 − 1
≈ 0.40%

26
© 2014-2024 AnalystPrep.
LOS 1e: calculate and interpret major return measures and describe
their appropriate uses

Other Return Measures

Gross and Net Return

The gross return is what an asset manager earns before subtracting various costs such as

management fees, custody fees, taxes, and other administrative expenses. However, it does

account for trading costs such as commissions.

Gross return does not consider management or administrative costs. For this reason, it is a

suitable metric for assessing and comparing the investment expertise of asset managers.

Net return is a metric for how much an investment has earned for the investor. It considers all

administrative and management costs that reduce an investor’s return.

Pre-tax and After-tax Nominal Return

Unless otherwise stated, all returns are nominal pre-tax returns in general. Depending on the

jurisdiction, different rates apply to capital gains and income. Long-term and short-term taxes

may also be applied to capital gains.

The after-tax nominal return is determined by subtracting any tax deductions applied to

dividends, interest, and realized gains from the total return.

Real Returns

Returns are typically presented in nominal terms, which consist of three components: the real

risk-free return as compensation for postponing consumption, inflation as compensation for the

loss of purchasing power, and a risk premium. Real returns are useful in comparing returns over

different periods, given that inflation rates vary over time.

Recall the relationship between the nominal rate and the real rate:

27
© 2014-2024 AnalystPrep.
 (1+ Nominal Risk-free rate)=(1+Real risk free rate)(1+Inflation premium)

We can find the connection between nominal and real returns by considering the real risk-free

rate of return and the inflation premium. This relationship can be expressed as:

(1 + Real risk-free rate)(1 + Risk premium)

(1 + Real Return) =
1 + Inflation premium

Real returns become particularly useful when you want to compare returns across various time

periods and different countries. This is especially important when returns are shown in local

currencies and when inflation rates vary from one country to another.

After-tax real return is the amount the investor receives as payment for delaying consumption

and taking on risk after paying taxes on investment.

Leveraged Returns

If an investor uses derivative instruments within a portfolio or borrows money to invest, then

leverage is introduced into the portfolio. The leverage amplifies the returns on the investor's

capital, both upwards and downwards.

The leveraged return considers the actual return on the investment and the cost of the borrowed

money. The cost of borrowing and financing fees are subtracted from the overall return produced

by the investment to determine the leveraged return.

Using the borrowed capital (debt) increases the size of the leveraged position by the additional

borrowed capital.

Intuitively, the leveraged return is given by:

Portfolio return
RL =
Portfolio equity
[R P × (VE + VB ) − (V B × rD )]
=
VE
VB
= RP + (R P − rD )
VE

28
© 2014-2024 AnalystPrep.
Where:

R L = Return earned on the leveraged portfolio.

R P = Total investment return earned on the leveraged portfolio.

VB = Value of debt in the portfolio.

VE = Value of equity in the portfolio.

rD = Borrowing cost on debt.

Example: Calculating Leveraged Return

For a $250,000 equity portfolio with an annual 9% total investment return, 40% financed by debt

at 6%, the leveraged return would be:

VB $100,000
RL = R P + (R P − rD ) = 9% + (9% − 6%) = 11%
VE $150,000

29
© 2014-2024 AnalystPrep.
Question

A $7,500,000 equity portfolio is 35% financed by debt at a cost of 5% per annum. If

the equity portfolio generates a 9% annual total investment return, the leverage

return is closest to:

A. 11.15%.

B. 14.00%.

C. 8. 25%.

The correct answer is A.

VB
R L = RP + (R P − rD )
VE
$2, 625, 000
= 9% + (9% − 5%) = 11.15%
$4, 875, 000

30
© 2014-2024 AnalystPrep.
 Learning Module 2: The Time Value of Money in Finance

LOS 2a: calculate and interpret the present value (PV) of fixed-income
and equity instruments based on expected future cash flows

The time value of money (TVM) is a fundamental financial concept. It emphasizes that a sum of

money is worth more in the present than in the future. There are three key reasons supporting

this principle:

The concept of opportunity cost suggests that money available today can be invested

and generate interest, increasing its value over time. By delaying the use of money, one

forgoes potential investment opportunities and the growth they offer.

Inflation poses a threat to the purchasing power of money in the future. Due to

inflation, the same amount of money may buy fewer goods or services in the future

compared to its present value. Consequently, having money now is advantageous since

its purchasing power diminishes as time progresses.

There is an element of uncertainty regarding future cash flows. Unexpected events

or circumstances may prevent the receipt of money as planned, rendering it less

reliable. Until the money is obtained, there is a level of uncertainty attached to its

availability and utility.

Time value of money calculations allow us to establish a given amount's future value.

Key Components of Time Value of Money

Discount rate or interest rate: The rate of discounting or compounding that you

apply to an amount of money to calculate its present or future value.

Time periods: The whole number of time periods over which a sum's present or future

value is calculated. These periods can be annually, semi-annually, quarterly, monthly,

weekly, etc.

Present value (PV): The amount of money you have today (or at time T = 0) is

31
© 2014-2024 AnalystPrep.
 referred to as the present value.

Future value (FV): The accumulated amount of money you get after investing the

original sum at a specific interest rate and for a given time period, say, two years.

Fundamental Formulas in Time Value of Money Calculations:

Let,

F V = Future value.

P V = Present value.

r = Stated discount rate per period.

N = Number of periods (Years).

Then the future value (F V ) of an investment is given by:

F V =P V (1 + r)N

If N is large such that N → ∞ the initial cashflow is compounded continuously:

F V = P V erN

To find the present value of the investment, we rewrite the above formula so that:

PV = F V (1 + r)−N

And for the continuous compounding, we have,

P V = F Vte−r N

Example: Calculating the Present Value of continuously Compounded Cashflows

A fund continuously accumulates to $4,000 over ten years at a 10% annual interest rate.

Calculate the closest present value of this fund.

32
© 2014-2024 AnalystPrep.
Solution

From the question, F V =4,000, rs =10%, N =10

So,

P V = F V e−N rs = $4, 000 × e−10×0.1 = $1 , 471.5178

Frequency of Compounding

When the frequency of compounding is more than once per year (quarterly, monthly, etc.), the

formulas are analogously defined as:

rs mN
F V N = P V (1 + )
m

Where:

m = Number of compounding periods per year.

N = Number of years.

rs = Annual stated rate of interest.

Intuitively, the formula for the P V is given by:

r s −m N
P V = F V (1 + )
m

In the following discussion, we shall let t = mN denote the number of compounding periods and
rs
m
= r denote the stated discount rate per period.

Calculation using a Financial Calculator

For calculating F V and PV using the BA II PlusTM Financial Calculator, use the following keys:

N = Number of compounding periods.

I /Y = Rate per period.

33
© 2014-2024 AnalystPrep.
P V = Present value.

F V = Future value.

P MT = Payment.

CP T = Compute.

It is important to note that the sign of P V and F V will be opposite. For example, if PV is

negative, then F V will be positive. Generally, an inflow is entered with a positive sign, while an

outflow is entered as a negative sign in the calculator.

Time Value of Money in Fixed-income Instruments

Fixed-income instruments are debt securities where an issuer borrows money from an investor

(lender) in exchange for a promised future payment. Examples of fixed-income instruments are

bonds, loans, and notes.

The market discount rate for fixed-income instruments, also known as yield-to-maturity (YTM), is

the interest rate investors require to invest in a specific instrument.

Cash Flow Patterns Associated with Fixed-Income Instruments

Cash flows in fixed-income instruments follow three general patterns: discount, periodic interest,

and level payments.

Discount Cash Flow Patterns

For discount cashflow patterns, an investor pays an initial discounted price (P V ) for the

instrument (such as a bond or a loan) and gets one payment (F V ) at the end maturity. The

investor's return is the interest earned, that is, the difference between the initial price and

principle (F V − P V ).

Discount bonds are also called zero-coupon bonds, and they do not have periodic interest

payments.

34
© 2014-2024 AnalystPrep.
The price of a discount bond can be calculated using the formula for the present value (P V ) of a

single cash flow, which is as follows:

P V = F Vt(1 + r)−t

Where:

F V = Future value.

P V = Present value.

r = Stated discount rate per period.

t = Number of compounding periods.

Example: Calculating the Future Value of a Zero-Coupon Bond

Assume Chad invests $8,000 in a zero-coupon bond that yields 8% annually and matures in four

years. The maturity value of this bond is closest to:

35
© 2014-2024 AnalystPrep.
Solution

Recall that:

F V = PV (1 + r)t

In this case, we have P V =8,000, r =8%, t=4 so that:

F V = 8 , 000(1 + 8%)4 = 10 , 883.91

Using the BA II Plus TM Financial Calculator

Steps Explanation Display

[2nd][QUIT] Return to standard calc Mode 0
[2nd][CLR TVM] Clears TVM Worksheet 0
2[N] Years/periods N=4
10[1/Y ] Set interest rate P V = −8, 000
0[P MT] Set payment P MT = 0
[CP T ][F V ] Compute future value F V = 10, 883.91

Note that zero-coupon bonds can be issued at negative interest rates. In this case, the price (PV)

of the bond is higher than the face value (FV).

Example: Calculating the Price of a Discount Bond Issued at Negative Interest Rates

In January 2018, the Swiss government issued 15-year sovereign bonds at a negative yield of

-0.08%. The present value (PV) of the bond per CHF100 of principal (FV) at the time of issuance

is closet to:

Solution

Recall for a zero-coupon bond,

PV = F V t(1 + r)−t
= 100(1 − 0.0008)−15 = 101.21

Periodic Interest Pattern (Coupon Instrument)

36
© 2014-2024 AnalystPrep.
A coupon instrument is a fixed-income investment. It includes periodic cash flows called coupons

and repays the principal at maturity. People often use these in coupon bond investments. These

instruments have a set schedule with regular, equal payments.

The pricing of a coupon bond involves calculating its present value (PV) based on the market

discount rate. The general formula for calculating the bond's price is derived from the

discounted cash flow model. It considers the coupon payments (PMTs) and the final principal

payment (FV) at maturity. The bond's price is determined by discounting each cash flow using

the market discount rate (r).

The formula used to calculate the present value (PV) of a coupon bond is as follows:

P MT PMT (P MT N + F VN )
PV(Coupon Bond) = + +. . . +
(1 + r) 1 (1 + r) 2 (1 + r)N

Where:

P MT = Coupon payment.

37
© 2014-2024 AnalystPrep.
F V = Future value.

r = Market discount rate (YTM).

N = Number of periods.

Example 1: Pricing a Coupon Bond on an Annual Basis

Suppose we have a 5-year bond with a face value of $1,000 and an annual coupon rate of 5%.

The market discount rate is 6%. The bond’s price is closest to:

Solution

P MT P MT (PMT N + F VN )
PV(Coupon Bond) = + +⋯+
(1 + r)1 (1 + r) 2 (1 + r)N

In this case, we have P MT =5% of $1,000=$50, r=6%, N =5 years, F V =$1,000 so that:

$50 $50 $50 $50

PV = + + +
1
(1 + 0.06) (1 + 0.06) 2 (1 + 0.06)3 (1 + 0.06)4
($50 + $1, 000)
+
(1 + 0.06)5
PV = $47.17 + $44.50 + $41.98 + $39.60 + $784.62 = $957.88

Therefore, the price of the bond would be $957.88

You could use a BA II Plus Calculator to solve the above question:

Steps Explanation Display

[2nd][QUIT] Return to standard calc Mode 0
[2nd][CLR TVM] Clears TVM Worksheet 0
5[N ] Years/periods N=5
6[1/Y ] Set interest rate I /Y = 6.00
50[PMT ] Set payment P MT = 50.00
1000[F V ] Set the face value F V = 1000.00
[CP T ][P V ] Compute the present value P V = −957.88

Example 2: Pricing a Coupon Bond With a Single Cash Flow on a semi-annual Basis

Assume an investor has a 2-year bond with a face value of $1000 and an annual coupon rate of

38
© 2014-2024 AnalystPrep.
6%, paid semi-annually. The market discount rate is 5%. The price of the bond is closest to:

Solution

Recall that:

P MT PMT (P MT N + F VN )
PV(Coupon Bond) = + +. . . +
(1 + r)1 (1 + r)2 (1 + r)N

Where:

6%
P MT = Coupon payments ($1, 000 × ) = $30 in this case.
2

F V = Future value ($1,000 in this case).

(5%)
r = Market discount rate (YTM), ( 2
= 2.5%) per period in this case.

N = Number of periods (4 periods in this case).

Plugging these values into the formula, we get:

$30 $30 $30 ($30 + $1, 000)

PV = + + +
(1.025)1 (1.025) 2 (1.025) 3 (1.025)4
P V = $29.27 + $28.55 + 27.85 + $933.13 = $1, 018.81

Therefore, the bond's price is $1, 018.81

You can easily use the BA II Plus calculator (or any other allowed financial calculator) to solve

the above question.

Steps Explanation Display

[2nd][QUIT] Return to standard calc Mode 0
[2nd][CLR TVM] Clears TVM Worksheet 0
4[N] Years/periods N=4
2.5[1/Y ] Set interest rate I /Y = 2.50
30[P MT ] Set payment P MT = 30.00
1000[F V ] Set the face value F V = 1000.00
[CP T ][P V ] Compute the present value P V = −1, 018.81

Perpetual Bonds

39
© 2014-2024 AnalystPrep.
Perpetual bonds are rare types of coupon bonds that do not have a stated date of maturity. They

are generally issued by firms seeking equity-like financing and usually include redemption

provisions.

The formula present value of perpetual bonds is obtained as follows: As N → ∞, the formula for

calculating PV of coupon changes as follows:

PV (perpetual bond)
lim PMT P MT (P MT N + F VN )
= [ + +⋯ + ]
(N → ∞) (1 + r) 1 (1 + r) 2 (1 + r)N
P MT
=
r

So, the present value of a perpetuity is given by:

P MT
PV =
r

Example: Perpetual Bond

In 2021, XYZ Financial (the holding company for XYZ Bank) issued $500 million in perpetual

bonds with a 4.00 percent semi-annual coupon. Calculate the bond's yield to maturity (YTM) if

the market price was $98.50 (per $100).

Solution

Recall,

P MT
PV =
r

Hence,

P MT
r=
PV

To solve this problem, we first need to calculate the semi-annual coupon payment, which is,

$100 × 4%
40
© 2014-2024 AnalystPrep.
 $100 × 4%
PMT(semi-annual coupon payment) = = $2 , PV = $98.50
2

Therefore,

$2
r= = 0.0203 = 2.03%
$98.50

The annualized yield-to-maturity is:

r = 0.0203 × 2 ≈ 4.06%

Level Payments (Annuity Instruments) Patterns

An annuity is a finite series of cash flows, all with the same value. A fixed-income instrument

with annuity payments provides a stream of periodic equal cash inflows over a finite period.

Level payments consist of interest and principal. Fixed income instruments with level payments

include fully amortizing loans such as mortgages.

There are two types of annuities: ordinary annuities and annuities due. Annuity due is a type of

annuity where payments start immediately at the beginning of time, at time t = 0. In other

words, payments are made at the beginning of each period.

On the other hand, an ordinary annuity is an annuity where the cashflows occur at the end of

each period. Such payments are said to be made in arrears (beginning at time t = 1). We shall

consider ordinary annuity in this section.

Ordinary Annuity

Remember that in an ordinary annuity, the series of payments does not begin immediately.

Instead, payments are made at the end of each period. It is further worth noting that the present

value of an annuity is equal to the sum of the current value of each annuity payment:

P V = A(1 + r)−1 + A(1 + r)−2 + .. . +A(1 + r)−N −1 + A(1 + r)−N

= A(1 + r)−1 + (1 + r)−2 +. . . +(1 + r)−(N−1) + (1 + r)−N
1 − (1 + r)N
PV = A
r

41
© 2014-2024 AnalystPrep.
Where:

A = Periodic cash flow.

r = Market interest rate per.

P V = Present value/ Principal Amount of the loan or bond.

N = Number of payment periods.

The periodic annuity is calculated as follows:

r(P V )
A=
(1 − (1 + r)−N

Consider a fully amortizing mortgage loan. In this case, the borrower receives the mortgage loan

now and promises to make periodic payments equal to the sum of interest and principal

payments.

42
© 2014-2024 AnalystPrep.
Note that the periodic mortgage payment is constant, but the proportion of the interest payment

decreases while the principal payment increases.

The cash flow pattern of a fully amortizing mortgage follows the pattern of an ordinary annuity

with a series of equal cash flows. As such, the periodic annuity (periodic payment) of a fully

amortizing mortgage is given by:

r(P V )
A=
1 − (1 + r)−t

Where:

43
© 2014-2024 AnalystPrep.
A = Periodic cash flow.

r = Market interest rate per period.

P V = Present value or principal amount of loan or bond.

t = Number of payment periods.

Example: Calculating the Periodic Payment of a Mortgage

Jake is looking to secure a fixed-rate 25-year mortgage to finance 75% of the value of an

$800,000 residential property. If the annual interest rate on the mortgage is 4.5%, Jake's monthly

mortgage payment is close to:

Solution

Remember,

r(P V )
A=
1 − (1 + r)−t

Where:

A = Periodic cash flow.

r = Market interest rate per period.

P V = Present value/ Principal Amount of the loan or bond.

t = Number of payment periods.

In this case, we have:

4.5%
r = 0.375% (= )
12

N = 300 months(= 25 years × 12 months/year)

P V = $600, 000(= 75% × $800, 000)

Plugging these values into the formula, we get:

0.00375 × $600, 000

44
© 2014-2024 AnalystPrep.
 0.00375 × $600, 000
A= = $3, 334.995 ≈ $3, 335
1 − (1 + 0.00375)−300

Using a BA II Plus financial calculator:

Steps Explanation Display

[2nd][QUIT] Return to standard calc Mode 0
[2nd][CLR TVM] Clears TVM Worksheet 0
300[N ] Years/periods N = 300
0.375[1/Y ] Set interest rate I/Y = 0.375
−600, 000[P V ] Set the present value P V = −600, 000.00
of the mortgage
0[F V ] Set the face value F V = 0.00
[CP T ][P MT ] Compute the periodic payment PMT = 3,334.99

Time Value of Money in Equity Instruments

Equity investments, such as stocks, enable an investor to acquire a fractional share/ownership by

the issuing company. This gives investors the right to receive a share of the company's available

cash flows as dividends.

In the context of equity instruments, the time value of money (TVM) is used to discount expected

future cash flows to determine their present value. This allows investors to value the company

shares.

The present value of expected future cash flows is calculated using a discount rate, r, which

represents the expected rate of return on the investment.

Common Approaches for Valuing Equity Instruments

Valuing equity investments depends on dividend cashflows, which can take one of three forms:

constant dividends, constant dividend growth rate, or changing dividend growth rate.

1. Valuing Equity Instruments based on Constant Dividend: The Constant Dividends model

values stocks based on the assumption that dividends will remain constant over time. The

preferred or common share dividend cash flows are in the form of an infinite series that is valued

like perpetuity. The formula for the constant dividends model is as follows:

45
© 2014-2024 AnalystPrep.
 ∞
Dt Dt
P Vt = ∑ =
i=1 (1 + r) i r

Where: PV t = Present value at time t

Dt = Dividend payment at time t

r = Discount rate.

Example: Valuing Equity Instruments based on Constant Dividend

Assuming we have a preferred stock with a dividend payment of $5 per year. The discount rate is

8%. The present value of the stock is closest to:

Solution

Recall,

Dt
P Vt =
r

In this case, D =$5, r=8%, P V =?

So,

5
PV = = $62.5
0.08

This means that the present value of the stock is $62.5.

2. Valuing Equity Instruments Based on Constant Dividend Growth Rate: The constant

dividend growth model is a method used to estimate the value of a stock based on its future

dividends. This model assumes that dividends will grow at a constant rate (g) forever. To derive

the formula for this model, we start by considering that the present value of a stock is equal

to the sum of its future dividends, discounted by the required rate of return r . If dividends

are assumed to grow at a constant rate, then each future dividend can be calculated by

multiplying the previous dividend by (1 + g) .

46
© 2014-2024 AnalystPrep.
Let Dt represent the expected dividend in the next period. The present value of the stock can

then be expressed as:

Dt Dt(1 + g) Dt(1 + g)2

PV t = + + +…
(1 + r) (1 + r)2 (1 + r)3
∞ D (1 + g)
t
=∑ i
i=1 (1 + r)

(1+r )
This is an infinite geometric series with a common ratio of . Using the formula for the sum of
(1+g)

an infinite geometric series, we can simplify this equation to:

Dt(1 + g) Dt+1
P Vt = =
r −g r−g

Where:

P Vt = Present value at time t .

Dt+1 = Expected Dividend in the next period.

r = Required rate of return.

g = Constant growth rate.

r−g>0

Therefore, this is the formula for calculating the present value of a stock using the constant

dividend growth rate. This model can help estimate the value of a stock when its future dividends

are expected to grow steadily.

Example: Valuing Equity Instruments based on Constant Dividend Growth Rate

Suppose a stock currently pays an annual dividend of $2.00 per share. The required rate of

return for this stock is 10%, and the dividends are expected to grow at a constant rate of 5% per

year indefinitely. Using the constant dividend growth model, the present value of this stock is

closest to:

Solution

47
© 2014-2024 AnalystPrep.
Recall that,

Dt(1 + g) Dt+1
P Vt = =
r −g r−g

In this case, we know that Dt =$2.00, r=10%, g=5%

So,

2 × 1.05 2.10
PV = = = $42
r −g 0.10 − 0.05

Therefore, the present value of the stock is $42

3. Valuing Equity Instruments with Changing Dividend Growth Rates is a dynamic

process. It begins with the investor buying a stock at an initial price and getting an initial

dividend. The unique aspect is that the dividend is expected to grow at a rate that evolves as the

company matures and shifts from high growth to slower growth. This valuation doesn't have a

single formula because it relies on assumptions about future dividend growth. However, a

common method is to use a multi-stage dividend discount model. This model assumes that

dividends will grow at different rates during various stages of the company's growth. To find the

stock's present value, you sum up the present values of dividends at each stage.

The Multi-Stage Dividend Discount Model builds on the Constant Dividend Growth Model. It

accommodates a company's transition from high initial growth to lower, more stable growth.

Let's say a company has a high short-term growth rate gs followed by a perpetual lower growth

rate gl . To find the present value (PV) of the stock at time t using this model, we compute it in

two stages:

I. First Part: The first part calculates the present value of dividends during the initial n

periods of higher growth (g s ). This is done by discounting the dividends for each period

by the required rate of return r using the following formula:

n Dt (1 + gs )i
P Vt = ∑
i =1 (1 + r)i

48
© 2014-2024 AnalystPrep.
 Where: PV = Present value. n = Number of [Link] = Dividend at time (t). gs = Initial

higher dividend growth rate.r = Required rate of return.

II. Second Part: The second part calculates the present value of dividends after the initial n

periods, assuming constant growth at a lower long-term rate (gl ). This can be simplified

using the geometric series simplification, where E(St + n) represents the terminal value

or stock value in n periods:

E(St + n)
P Vt =
(1 + r)n

D t+ n+1
Where: E(St + n) = and g l is the lower, more stable dividend growth rate.
( r−g l)

Example: Valuing Equity Instruments based on Changing Dividend Growth Rate

Assuming we have a stock with an expected dividend payment of $2 in one period, and the

discount rate is 10%. The stock is expected to have a high dividend growth rate of 20% for the

first three years, followed by a slower growth rate of 5% thereafter. Calculate the present value

of the stock.

Solution

First, we calculate the present value of the dividends during the high growth period:

Recall that,

n Dt (1 + gs )i
P Vt = ∑
i =1 (1 + r)i

In this case, Dt =$2, g s=0.20, r=0.10, n =3

So,

2 2 × (1 + 0.20)1 2 × (1 + 0.20)2
P V1 = + +
(1 + 0.10)1 1 + 0.10)2 1 + 0.10)3
P V1 = 1.818 + 1.983 + 2.163 = 5.965
P V1 = $5.97

49
© 2014-2024 AnalystPrep.
Next, we calculate the present value of the dividends during the slower growth period, assuming

that dividends will grow at a constant rate of 5% thereafter:

Recall that,

Dt+n+1
E(St + n) =
(r − gl )

So,

2(1 + 0.20)3 × (1 + 0.05) 3.629

E(St + n) = = = $72.578
0.10 − 0.05 0.05

Finally, we calculate the present value of P4

$72.578
P V2 = = $54.527
(1 + 0.10)3

The total present value of the stock is the sum of P V1 and P V2

P Vtotal = P V1 + P V2
= $5.965 + 54.527
= $60.493 ≈ $60.49

50
© 2014-2024 AnalystPrep.
Question

Five years ago, Milton Inc. issued corporate bonds with a 15-year maturity. The bonds

have a semi-annual coupon rate of 7.8% per annum, and the current yield to maturity

is 8.5% per annum. The current price of Milton Inc's bonds (per CAD100 of par value)

is closest to:

A. CAD91.23.

B. CAD95.35.

C. CAD96.15.

The correct answer is B.

Solution

Recall the formula for calculating the price of a bond:

P MT PMT (P MT N + F VN )
PV(Coupon Bond) = + +. . . +
(1 + r)1 (1 + r)2 (1 + r)N

First, let's get the semi-annual equivalent rates:

7.8%
The semi-annual coupon rate is 2 = 3.9%.

8.5%
The semi-annual yield to maturity is 2 = 4.25%.

Next, we find the number of periods remaining until the bond matures:

Since the bonds were issued 5 years ago and have a 15-year maturity, 10(= 15 − 5)

years remain to maturity. Since interest is paid semi-annually, this equates to

10 × 2 = 20 periods.

You can plug the above values into the general formula, consuming valuable time.

Therefore, using BA II plus a financial calculator,

51
© 2014-2024 AnalystPrep.
Steps Explanation Display
[2nd][QUIT] Return to standard calc Mode 0
[2nd][CLR TVM] Clears TVM Worksheet 0
20[N ] Years/periods N = 20
4.25[1/Y ] Set interest rate I /Y = 4.25
3.9[P MT ] Set the periodic P MT = 3.90
coupon payment
100[P V ] Set the face F V = 100.00
value of the bond
[CP T ][P V ] Compute the present value PV = −95.35

52
© 2014-2024 AnalystPrep.
LOS 2b: Calculate and interpret the implied return of fixed-income
instruments and the required return and implied growth of equity
instruments given the present value (PV) and cash flows

Implied Return for Fixed-Income Instruments

The growth rate is the rate at which the market expects an asset to grow. On the other hand,

implied return reflects a return based on the current price and future security cash flows.

Consider a fixed-income instrument. If we have its present value and assume all future cash

flows happen as expected, the discount rate, rrr, or yield-to-maturity, YTM, shows the implied

return under these assumptions for the cash flow pattern.

Now, take an equity investment. If we have the present value, future value, and discount rate, we

can find the implied growth rate that aligns with these values.

The implied return or growth rate provides a view of the market expectations incorporated into

an asset's market price. Understanding these expectations is critical for investors when making

investment decisions.

Calculating the Implied Return for Fixed-Income Instruments

Discount Bond

In the case of a discount bond or instrument, recall that an investor receives a single principal

cash flow (F V ) at maturity, with (F V − PV ) representing the implied return.

To solve for the implied return earned over the life of an instrument (N periods), we can

rearrange the single cash flow present value formula.

Recall that the single cash flow present value formula is:

P V = F Vt(1 + r)−t

53
© 2014-2024 AnalystPrep.
Where:

F V = Future value.

P V = Present value.

r = Stated discount rate per period.

t = Number of compounding periods.

To solve for r, we can rearrange this formula as follows:

1
F Vt F Vt t
r=√ t −1 = ( ) −1
PV PV

We use this formula to calculate the periodic return earned during the instrument's life (t

periods) based on its present value (or price) and future value.

Example: Calculating the Implied Return for a Discount Bond

Consider a zero-coupon bond with price of $900, a future value of $1,000, and a maturity of 5

years. Calculate the implied annualized return r.

Solution

Recall that,

1
F Vt t
r=( ) −1
PV

In this case, t =5, FV t=$1,000, P V =$900

So,

1
1000 5
r=( ) − 1 = 2.13%
900

This means that an investor who purchases this zero-coupon bond at a price of $900 and holds it

for five years would earn an annualized return of 2.13%.

54
© 2014-2024 AnalystPrep.
Coupon Bonds

Recall that fixed-income instruments that pay periodic interest have cash flows throughout their

life until maturity. The yield-to-maturity (YTM) is a single implied market discount rate for all

cash flows, regardless of timing. It assumes an investor expects to receive all promised cash

flows through maturity and reinvest any cash received at the same YTM.

The present value of a fixed-income instrument with periodic interest can be calculated using the

following formula:

P MT 1 P MT2 (P MTN + FV N )
PV = + +⋯+
(1 + r) 1 (1 + r)2 (1 + r)N

Where:

P V = Present value (or price) of the instrument.

P MT = Periodic payment.

F V = Bond's principal.

N = Number of periods to maturity.

r = Discount rate (or internal rate of return) (YTM).

Example: Implied Return for Fixed-income Instruments With Periodic Interest

Consider a five-year corporate bond issued in 2023 with a 4.00 percent annual coupon and a

price of USD110.00 per USD100 principal three years later. If Milka can reinvest periodic

interest at the original YTM of 4.00 percent, the implied three-year return is closest to:

Solution

We can calculate the future value (FV) after three years, including the future price of 110.00 and

all cash flows reinvested to that date:

F V3 = P MT1 (1 + r)2 + P MT2 (1 + r) + P MT3 + P V3

= 4 × (1.04)2 + 4 × (1.04) + 4 + 110.00
= $122.49

55
© 2014-2024 AnalystPrep.
We can then solve for Milka's annualized return, r, using the formula for implied return since we

have P V =100, F V =122.49, N = 3 as follows:

F Vt 122.49
r = √t − 1 = √3 − 1 = 3.65%
PV 110

This means that Milka, who purchased the corporate bond at a price of 100 and held it for three

years, would earn an annualized return of 3.65%.

Example: Calculating the Yield-to-Maturity of a Coupon Bond

CityGroup Corp. issued a corporate bond 7 years ago with a face value of $1,000 and a 20-year

maturity. The bond pays annual interest at a coupon rate of 6%. Currently, the bond is trading at

$1,120. The yield to maturity (YTM) of CityGroup Corp.'s bond is closest to:

Solution

Using the BA II Plus calculator, we solve the question as follows:

We have

Steps Explanation Display

[2nd][QUIT] Return to standard calc Mode 0
[2nd][CLR TVM] Clears TVM Worksheet 0
13[N] Years/periods N = 13
−1, 120[PV ] Set the present value P V = −1, 120
of the bond
60[PMT ] Set the periodic P MT = 3.90
coupon payment
1, 000[P V ] Set the face value F V = 1000.00
of the bond
[CP T ][I /Y ] Compute the YTM I /Y = 4.74%

Therefore, the YTM is 4.74%.

Implied Return and Growth for Equity Instruments

The value of a stock is determined by both the expected return and the growth of its cash flows.

56
© 2014-2024 AnalystPrep.
By assuming a constant growth rate for dividends, we can use the formula for the present value

of an equity investment to calculate the stock's implied return or growth rate.

Implied Return

Recall that the present value of a stock for constant growth of dividends is given by:

Dt(1 + g) Dt+1
P Vt = =
r −g r−g

Where:

P Vt = Present value at time t .

Dt = Expected Dividend in the next period.

r = Required rate of return.

g = Constant growth rate.

r−g>0

Therefore, we can calculate the implied return on a stock given its expected dividend yield and

implied growth by rearranging the above formula as follows:

Dt (1 + g) Dt+1
r= +g = +g
P Vt PV t

In simple terms, if we assume a stock's dividends will grow at a steady rate forever, the implied

return is the combination of its expected dividend yield and the constant growth rate.

Example: Implied Return and Growth

Assume Apple Inc. stock is trading at a share price of USD150.00, and its annualized expected

dividend per share during the next year is USD2.00.

Moh, an analyst, projects that Apple's dividend per share will increase at a constant rate of 5%

per year indefinitely. The required return expected by investors on the stock is closest to:

57
© 2014-2024 AnalystPrep.
Solution

Recall that the implied return formula is,

Dt (1 + g) Dt+1
r= +g = +g
P Vt PV t

In this case, D =$2.00, P V =$150, g=5%

Therefore,

2.00(1.05)
r= + 0.05 = 6.4%
150

Implied Growth

We can also solve for a stock's implied growth rate, which is given by the following formula:

r × P Vt − Dt r − Dt+1
g= + Dt =
P Vt P Vt

Example: Calculating the Implied

Consider the previous example. Suppose Moh believes that Apple stock investors should expect a

return of 8%, calculate the implied dividend growth rate for Apple Inc.

Solution

Recall that the formula for calculating implied growth is as follows.

r × P Vt − Dt r − Dt+1
g= + Dt =
P Vt P Vt

So,

2.00 × 1.05
g = 0.08 − = 0.066 = 6.60%
150

Price-to-Earnings Ratio

58
© 2014-2024 AnalystPrep.
In equity instruments, it is common practice to compare the price-to-earnings ratio.

The price-to-earnings (P/E) ratio is a valuation metric that compares the current share price

of a stock to its earnings per share. Investors and analysts use it to determine the relative value

of a company’s shares compared to other companies or the market.

A stock with a higher price-to-earnings ratio is more expensive than a lower one, as investors are

willing to pay more for each unit of earnings. This ratio is also a valuation metric for stock

indexes, such as S&P 500.

Relating P/E Ratio to Expected Future Cash Flows

The P/E ratio can relate to our earlier discussion on a stock's price (PV) to the expected future

cash flow relationship. Recall the following equation:

D t × (1 + g)
P Vt =
r−g

By dividing both sides of the equation by E t, which represents earnings per share for period t , we

get the following equation:

Dt
× (1 − g)
PV t Et
=
Et r−g

Where:

P Vt
= Price-to-earnings (P/E) ratio.
Et

Dt
= Dividend payout ratio.
Et

g = Growth rate.

r = Required rate of return.

The dividend payout ratio represents the percentage of a company's earnings paid out to

shareholders in the form of dividends.

59
© 2014-2024 AnalystPrep.
Typically, the forward P/E ratio, which is based on a projection of a company's earnings per

share for the next period (t + 1), is used. This ratio is positively correlated with higher expected

dividend payouts and growth rates but negatively correlated with the required return.

Therefore, the equation:

Dt
× (1 − g)
PV t Et
=
Et r−g

This can be simplified as below to find the forward P/E ratio:

D t+1
P Vt E t+1 Dt+1 1
= = ×
Et+1 r−g Et+1 r −g

Example: Solving for Implied Dividend Growth Rate

Suppose a company has a forward P/E ratio of 15, a dividend payout ratio of 40%, and a required

return of 10%. The implied dividend growth rate for this company is closest to:

Solution:

First, we can use the formula for the forward P/E ratio to solve for the implied dividend growth

rate:

P Vt Dt+1 1
= ×
Et+1 Et+1 r −g

Where:

P Vt = Present value at time t .

Et+1 = Earnings per share for the next period.

Dt+1= Dividend payout for the next period.

r = Required return.

g = Implied dividend growth rate.

60
© 2014-2024 AnalystPrep.
Substituting the given values into the formula, we get:

0.4
15 =
0.1 − g

Solving for g, we get:

0.4
g = 0.1 − = 0.0733
15

Therefore, the implied dividend growth rate for this company is 7.33%.

Example: Solving for Required Return

Let's assume you are not given the required rate of return in the question above so that the

company has a forward P/E ratio of 15, a dividend payout ratio of 40%, and an implied dividend

growth rate of 7.33%. What is the required return for this company?

Solution

Recall the formula for the forward P/E ratio to solve for the required return:

P Vt Dt+1 1
= ×
Et+1 Et+1 r −g

Substituting the given values into the formula, we get:

0.4
15 =
r − 0.0733

Solving for r, we get:

0.4
r= + 0.0733 = 0.1000
15

Therefore, the required return for this company is 10%.

61
© 2014-2024 AnalystPrep.
Question

Edmund company's stock trades at USD50.00. The company pays an annual dividend

to its shareholders, and its most recent payment of USD 2.00 occurred yesterday.

Analysts following the company expect its dividend to grow at a constant rate of 4

percent per year. What is the company's required return?

A. 8.16%.

B. 8.48%.

C. 9.16%.

The correct answer is A.

Solution

Recall that:

Dt+1 × (1 + g)
PV =
r−g

Where:

P V = Current stock price.

Dt+1 = Recent dividend payout.

g = Expected dividend growth rate.

r = Required return.

Substituting the given value into the formula, we get:

2 × (1 + 0.04)
50 =
r − 0.04

Solving for r, we get:

2 × (1 + 0.04)
62
© 2014-2024 AnalystPrep.
 2 × (1 + 0.04)
r= + 0.04 = 0.0816
50
r = 0.0816

Edmund's required return is 8.16%.

63
© 2014-2024 AnalystPrep.
LOS 2c: explain the cash flow additivity principle, its importance for the
no-arbitrage condition, and its use in calculating implied forward
interest rates, forward exchange rates, and option values

A timeline is a physical illustration of the amounts and timing of cashflows associated with an

investment project. For cashflows that are regular and of equal amounts, the standard annuity

formula or the financial calculator can be used. However, a timeline is preferred for irregular,

unequal, or both cashflows.

Remember that the general formula that relates the present value and the future value of an

investment is given by:

F VN = P V (1 + r)N

Where:

P V = Present value of the investment.

F VN = Future value of the investment N periods from today.

r = Rate of interest per period.

We can represent this in a timeline:

64
© 2014-2024 AnalystPrep.
In a particular timeline, a time index, t, represents a particular point in time, a specified number

of periods from today. Therefore, the present value is the investment amount today (t = 0), and

by using this amount, we can calculate the future value (t = N). Alternatively, we can use the

future value to calculate the present value.

The above argument can be written in terms of the present value. That is:

P V = F VN (1 + r)−N

Example: Applying a Timeline to Model Cashflows

A fixed-income investor receives a series of payments, each amounting to $6,500, set to be

received in perpetuity. Payments are to be made at the end of each year, starting at the end of

year 4. If the discount rate is 9%, then what is the present value of the perpetuity at t = 0?

Solution

We would then draw a timeline to understand the problem better:

65
© 2014-2024 AnalystPrep.
Here, we can see that the investor is receiving $6,500 in perpetuity. Recall that the PV of

aperpetuity is given by:

C
PV of a perpetuity =
r

So, in this case:

$6, 500
PV 3 = = $72, 222
9%

This is the value of the perpetuity at t = 3, so we need to discount it for three more periods to

get the value at t = 0. Using the formula:

P V0 = F V N (1 + r)−N

$72, 222
The PV at time zero is = $55 , 769
(1+0.09)3

Why is the Use of a Timeline Recommended?

There are many instances in real life when cashflows are uneven. A good example is a pension

66
© 2014-2024 AnalystPrep.
contribution that varies with age. Applying one of the basic time value formulae is impossible in

such cases. You are advised to draw a timeline even if the question appears relatively

straightforward. It will help you understand the question structure better. A timeline also helps

candidates add cashflows indexed to the same period and apply the value additivity principle.

Cashflow Additivity Principle

According to the cashflow additivity principle, the present value of any stream of cashflows

indexed at the same point equals the sum of the present values of the cashflows. This principle

has different applications in time value of money problems. Besides, this principle can be applied

to in different economic scenarios.

Application of Cashflow Additivity

Investing in Different Currencies

The principle of cash flow additivity can be applied to scenarios involving different currencies by

converting all cash flows to a common currency using the appropriate exchange rates. Doing so

allows us to compare and combine cash flows from different currencies and make investment

decisions based on their combined value.

For example, suppose we have two investment opportunities, one in US dollars and one in Euros.

We can convert the expected cash flows from the Euros investment into US dollars using the

appropriate exchange rates. Then, we can compare the combined value of the two investments

and decide based on their relative values.

Dealing with different currencies assumes continuous compounding. Recall that the present

value of a continuous compounding is given by:

PV = F VN e−N r s

Example: Applying the Cash Flow Additivity Principle

Consider an investor with USD 2,000 who wants to invest it for three months. The investor can

67
© 2014-2024 AnalystPrep.
choose between two options: investing in the US government debt or German government debt.

Option 1: Investing in US Government Debt

The investor can invest his USD 2,000 in a three-month US Treasury bill. This means that he

lends the government USD 2,000, and it promises to pay him back with interest in three months.

The interest is 3%, so after three months, he will receive:

Recall that,

3
F V = 2 , 000 × e0.03 × 12 = USD 2 , 015

Option 2: Investing in German Government Debt

The investor chooses to invest in German government debt. To do this, the investor must convert

his USD 2,000 into Euros at the current exchange of EUR/USD = 0.92 (1 USD =0.92EUR). This

means that the investor will receive (2, 000 × 0.92) = EUR 1840 . He can then lend this money to

the German government by investing in a three-month German Treasury bill. Assuming the

interest rate is 0.06 percent, after three months, the investor will receive:

0.06 × 3
F V = EUR 1840 × e 12

= EUR 1, 867.81

Assuming the investor wants his money in US dollars, we need to convert the EUR1867.81 back

into USD at the forward exchange rate of USD/EUR = 1.0788. This means that the investor will

receive:

EUR 1,867.81 ×USD 1.0788

1 EUR
= $2, 014.99 ≈ 2, 015

Both options give you the same amount of money after three months: USD 2,015. The difference

is that one option involves investing in US dollars, and the other involves converting your money

into Euros and back into US dollars.

The forward exchange rate of 1.0788 USD/EUR is important because it determines how much

money you, the investor, will receive when converting your Euros back into US dollars. If this

rate differs from 1.0788, there would be an arbitrage opportunity in converting Euros to dollars.

68
© 2014-2024 AnalystPrep.
More on foreign exchange rates will be discussed later in the curriculum.

Implied Forward Rates

Consider two zero-coupon bonds. Bond A has a maturity of two years and a yield of 2% per

annum, while bond B has a maturity of four years and a yield of 3%. An investor, who doesn't

seek to take advantage of price differences and is risk-neutral, has $1,000 to invest. The investor

has two investment options that earn the same return.

Option 1: The investor can put their money into bond B now, which has an annual yield of 3%,

and will pay out at the end of the four years. The Future Value (FV) of this investment in four

years, using the formula for compound interest, is:

F V4 = P V0 (1 + r4 )4 = 1000(1.03)4 = 1, 125.51

Option 2: Alternatively, the investor can initially invest in bond A and, after two years, reinvest

the proceeds at a forward rate F2 ,2 which represents a two-year forward rate starting in year

two.

By the principle of cash flow additivity, a risk-neutral investor will not prefer one option over the

other - they are indifferent between Options 1 and 2. This is because the Future Values of both

investments at the end of four years should be the same:

F V4 = P V0 (1 + r4 )4 = PV 0 (1 + r2 )(1 + F2 ,2 )

In this scenario, this simplifies to:

1, 125.51 = 1, 000(1.02)2 (1 + F2, 2)

Solving this equation for the forward rate gives:

1, 125.51
⇒ F2,2 = − 1 = 8.18%
1 , 000(1.02)2

Therefore, to prevent arbitrage opportunities, the forward rate F2 ,2 should be set to 8.18%. This

69
© 2014-2024 AnalystPrep.
ensures that there is no potential for risk-free profits, maintaining market efficiency.

Cashflow Additivity and Option Pricing

Cash flow additivity can be used to determine the fair price of an option contract. An option

contract gives the buyer the right, but not the obligation, to buy (call) or sell (put) an underlying

asset at a specified price within a certain period.

Cash flow additivity allows investors to compare different strategies and determine a no-

arbitrage price for a financial instrument.

Example: Illustrating Cashflow Additivity in Call Option

Consider a stock that costs $100 now. Its price might increase by 30% to $130 or decrease by

15% to $85 in one year.

Let's say an investor wants to sell a call on the stock that gives the buyer the right, but not the

obligation, to purchase the asset for $120. The principle of cash flow additivity can be used to

70
© 2014-2024 AnalystPrep.
determine the contract's no-arbitrage price.

If the stock price goes up, the contract is worth cu1 = 10. That's because the buyer can use the

contract to buy the asset for $120 and then sell it for $130, making a profit of $10. But if the

price goes down, the contract is worth nothing. The buyer wouldn't want to use the contract to

buy the asset for $120 when they could just buy it for $85 without the contract.

The underlying argument here is that the value of the option is each movement of the stock

option may be used to construct a risk-free portfolio (the value of the portfolio is the same in

both scenarios).

Denote the initial value of the call option by c0 , which we wish to determine using cash flow

additivity and no-arbitrage pricing. Also, denote the value of the portfolio at t = 0 by V0 , when

the stock price increases by V1u and when the stock price decreases by V1d

Assume that at t = 0 creates a risk-free portfolio by selling a call option at c0 and buying 0.22

units of underlying assets. Then, the value of the portfolio at inception is:

V0 = 0.22 × 100 − c 0

This portfolio is called a replicating portfolio because it is designed to create a matching

future cash flow stream to that of a risk-free asset.

Similarly, in each scenario of stock price decrease and increase:

V 1u = 0.22 × 130 − 10 = 18.89

V 1d = 0.22 × 85 − 0 = 18.89

Intuitively, the value replicating portfolio equals 18.89, whether the stock prices rise or decline.

As such, the replicating portfolio is risk-free and can be discounted as a risk-free asset. Assuming

that risk-free rate is r = 2.5%, then:

V 0 = V 1u (1 + r)−1 = V1d (1 + r)−1

= 18.89(1.025)−1 = 18.43

At this point, we can calculate the value of c 0 rearranging the initial portfolio value equation:

71
© 2014-2024 AnalystPrep.
 V0 = 0.22 × 100 − c0
⇒ 18.43 = 0.22 × 100 − c0
∴ c 0 = 3.57

As such, the fair price of the call option is $3.57, which the seller expects to receive from the

buyer.

72
© 2014-2024 AnalystPrep.
Question

The current USD/CHF exchange rate is 0.9. The risk-free interest rates for one year

are 2% for the US dollar and 1% for the Swiss franc. Which of the following one-year

USD/CHF forward rates would best prevent arbitrage opportunities?

A. USD/CHF 0.909.

B. USD/CHF 0.099.

C. USD/CHF 1.122.

Solution

The correct Answer is C.

Dealing with different currencies assumes continuous compounding. Recall that the

future value of a continuous compounding is given by:

F V = P VN eNr s

So,

In one year, a single unit of Swiss franc invested is:

e0.01 = CHF 1.0101

In one year, a single unit of Swiss Franc converted to US dollars and then invested

risk-free is worth;

1
1 CHF = ≈ 1.1111 USD
0.9

F VU SD = 1.1111.e0.02 ≈ 1.1334 USD

Therefore, to convert USD 1.1334 into CHF 1.0101 requires a forward exchange rate

of:

1.1334
73
© 2014-2024 AnalystPrep.
1.1334
= USD/CHF 1.1221
1.0101

74
© 2014-2024 AnalystPrep.
 Learning Module 3: Statistical Measures of Asset Returns

LOS 3a: calculate, interpret, and evaluate measures of central tendency

and location to address an investment problem

The center of any data is identified via a measure of central tendency. A measure of central

tendency for a series of returns reveals the center of the empirical distribution of returns. They

include mean, mode, and median.

Measures of location help us understand where data points tend to cluster. These measures

include central tendency measures such as mean, median, and mode. There are also other

measures that provide different insights into how the data is spread out or located within a

distribution.

Measures of Central Tendency

Arithmetic Mean

The arithmetic mean is the sum of the values of the observations in a dataset divided by the

75
© 2014-2024 AnalystPrep.
number of observations.

Recall the formula: denoted by R̄i arithmetic mean for an asset i is a simple process of finding the

average holding period returns. It is given by:

Ri, 1 + R i,1 + … + Ri ,T−1 + R iT 1 T

R̄ i = = ∑ Rit
T T t=1

Where:

R it = Return of asset i in period t.

T = Total number of periods.

For example, if a share has returned 15%, 10%, 12%,

and 3% over the last four years, then the arithmetic
mean is computed as follows:

1 T 1
R̄ i = ∑ R it = (15% + 10% + 12% + 3%) = 10%
T t=1 4

Population Mean and Sample Mean

The population mean is the summation of all the observed values in the population, ∑ Xi divided

by the total number of observations, N . The population mean differs from the sample mean,

which is based on a few observed values chosen from the population. Thus:

∑ Xi
Population mean =
N
∑ Xi
Sample mean =
n

Analysts use the sample mean to estimate the actual population mean.

The population mean and the sample mean are both arithmetic means. The arithmetic mean for

any data set is unique and is computed using all the data values. Among all the measures of

central tendency, it is the only measure for which the sum of the deviations from the mean is

76
© 2014-2024 AnalystPrep.
zero.

Example: Calculating the Arithmetic Mean

The following are the annual returns realized from a

given asset between 2005 and 2015.
{12% 13% 11.5% 14% 9.8% 17% 16.1%
13% 11% 14%}
Calculate the population mean.
Solution
0.12 + 0.13 + 0.115 + 0.14 + 0.098 + 0.17 + 0.161 + 0.13 + 0.11 + 0.14
Population mean =
10
= 0.1314 = 13.14%

Median

The median is the statistical value located at the center of a data set organized in ascending or

descending order.

Consider a sample of n observations. For an odd-numbered of observations, the median is the

( n+1)
observation that is located in position. On the other hand, if the number of observations is
2
n n+2
even, the media is the mean value of the observations located in and position.
2 2

Unlike the arithmetic mean, the median resists the effects of extreme observations. However, it

only gives the relative position of the ranked observations without considering all observations

relating to the size of the observations.

Example: Calculating the Median

The following are the annual returns on a given asset realized between 2005 and 2015.

{12% 13% 11.5% 14% 9.8% 17% 16.1% 13% 11% 14%}

77
© 2014-2024 AnalystPrep.
The median is closest to:

Solution

First, we arrange the returns in ascending order:

{9.8% 11% 11.5% 12% 13% 13% 14% 14% 16.1% 17%}

Since the number of observations is even, the median return will be the middle point of the two
n ( n +2)
middle values in the positions and .
2 2

n 10 ( n+2) 12
The value occupying = = 5th position is 13, and the value located in = = 6th
2 5 2 2

position is 13, so that the mode is:

13% + 13%
= 13%
2

Mode

The mode is the value that appears most often in a dataset. Sometimes, a dataset has a mode;

sometimes, it doesn't. If all the observations in a dataset are different and no value repeats more

than others, then the dataset has no mode.

A dataset with one mode is called unimodal. When there are two modes, it's called bimodal. If the

distribution has three frequently occurring values, it's termed trimodal.

An interval with the highest frequency is called the modal interval (or intervals) in a frequency

distribution. For instance, in the frequency distribution below, the modal interval is -1.0 to 0.0

with an absolute frequency of 520.

78
© 2014-2024 AnalystPrep.
 Return Bin Absolute Relative Cumulative Cumulative]
(%) Frequency Frequency(%) Absolute Relative
Frequency Frequency (%)
-6.0 to -5.0 2 0.16 2 0.16
-5.0 to -4.0 8 0.64 10 0.80
-4.0 to -3.0 27 2.16 37 2.96
-3.0 to -2.0 80 6.40 117 9.36
-2.0 to -1.0 485 38.80 602 48.16
-1.0 to 0.0 520 41.60 1 , 122 89.76
0.0 to 1.0 100 8.00 1 , 222 97.76
1.0 to 2.0 24 1.92 1 , 246 99.68
2.0 to 3.0 3 0.24 1 , 249 99.92
3.0 to 4.0 1 0.08 1 , 250 100.00

In a histogram, the modal interval always has the highest bar.

The mode is the only measure of central tendency that can be used with nominal data. Nominal

data refers to a type of data that is categorized into distinct categories or groups without any

inherent order or numerical value. Examples of nominal data include gender (male, female), eye

color (blue, brown, green), and marital status (single, married, divorced).

Example: Calculating the Mode

79
© 2014-2024 AnalystPrep.
Determine the mode from the following data set:

{20% 23% 20% 16% 21% 20% 16% 23% 25% 27% 20%}

Solution

The mode is 20%. It occurs four times, a frequency higher than any other value in the data set.

Clearly, this dataset is unimodal.

Dealing With Outliers

An outlier may represent a distinct value in a population. In addition, it may show that there was

an error in recording the value, or it was generated from a different population.

When working with a sample with outliers, we can potentially transform the variable or choose

another variable that achieves the same objective. If these observations prove to be impossible,

there are three options:

Option 1: Take no action and use the data as it is. If these observations are accurate, then this is

appropriate.

Option 2: Remove outliers using the trimmed mean. For example, when calculating the central

tendency, a 4 percent trimmed mean excludes the lowest 2% and highest 2% of values.

Option 3: Substitute a different value for the outliers. The winsorized mean is an illustration of a

central tendency that does this. For instance, when computing a 96% winsorized mean, the value

at or above the lowest and highest 2% is assigned the lowest and highest 2% values.

Measures of Location

Quartiles, quintiles, deciles, and percentiles are values or cut points that partition a finite

number of observations into nearly equal-sized subsets. The number of partitions depends on the

type of cut point involved.

80
© 2014-2024 AnalystPrep.
Quartiles

They divide data into four parts. The first quartile, Q1, is referred to as the lower quartile, and

the last quartile, Q4, is the upper quartile. Q1 splits the data into the lower 25% and upper 75%

values. Similarly, the upper quartile subdivides the data into the lower 75% of the values and the

upper 25%. The difference between the upper and lower quartiles is known as the interquartile

range, which indicates the spread of the middle 50% of the data.

Quintiles

Though rarely used in practice, quintiles split a set of data into five equal parts, i.e., fifths.

Therefore, the second quintile splits data into the lower 40% of the values and the upper 60%.

Deciles

The deciles subdivide data into ten equal parts. There are 10 deciles in any data set. For

example, the fourth decile splits data into the lower 40% of the values and the upper 60%.

Percentiles

Percentiles split data into 100 equal parts, i.e., hundredths. So, for instance, the 77th percentile

splits the data into the lower 77% of the values and the upper 23%.

Financial analysts commonly use the four types of subdivisions to rank investment performance.

You should note that quartiles, quintiles, and deciles can all be expressed as percentiles. For

instance, the first quartile is just the 25th percentile. Similarly, the fourth decile is simply the

40th percentile. This enables the application of the formula below.

(n + 1) y
Position of percentile =
100

Example: Calculating Quartiles

Given the following distribution of returns, calculate the lower quartile.

81
© 2014-2024 AnalystPrep.
{10% 23% 12% 21% 14% 17% 16% 11% 15% 19%}

Solution

First, we have to arrange the values in ascending order:

{10% 11% 12% 14% 15% 16% 17% 19% 21% 23%}

Next, we establish the position of the first quartile. This is simply the 25th percentile. Therefore:

(10 + 1) 25
P 25 = = 2.75 th value
100

Since the value is not straightforward, we have to extrapolate between the 2nd and the 3rd data

points. The 25th percentile is three-fourths (0.75) of the way from the 2nd data point (11%) to the

3rd data point (12%):

11% + 0.75 × (12 − 11) = 11.75%

Box and Whisker Plot

Box and whisker plot is used to display the dispersion of data across quartiles. A box and whisker

plot consists of a "box" with "whiskers" connected to the box. It shows the following five-number

summary of a set of data.

82
© 2014-2024 AnalystPrep.
Example: Box and Whisker Plot

Consider the following box and whisker plot:

1. Which of the following is most likely the median?

83
© 2014-2024 AnalystPrep.
 A. 10.

B. 10.5.

C. 11.

Solution

The correct answer is B.

(10 + 11)
Median or Quartile 2 (Q2) = = 10.5
2

2. Which of the following is most likely the interquartile range?

A. 4.5.

B. 6.5.

C. 11.

Solution

The correct answer is C.

Interquartile range is Q3 − Q1 = 15 − 4 = 11

3. Which of the following is most likely the 3rd quartile?

A. 4.

B. 15.

C. 18.

Solution

3 × (n + 1) 3 × (16 + 1)
Quartile 3 (Q3) = = = 12.75 th term, which is 14.75
4 4

84
© 2014-2024 AnalystPrep.
Quantiles in Investment Practice

Quantiles have two main purposes in investment practice. First, quantiles are used to rank

performance. Secondly, quantiles can be used in investment research for comparison purposes.

For example, companies can be clustered into deciles to compare the performance of small

companies with the large ones. In this case, the first decile will contain the portfolio of

companies with the smallest market values, while the tenth decile will contain the companies

with the largest market values.

85
© 2014-2024 AnalystPrep.
Question

A mutual fund achieved the following rates of growth over an 11-month period:

{3% 2% 7% 8% 2% 4% 3% 7.5% 7.2% 2.7% 2.09%}

The 5th decile from the data is closest to:

A. 2%.

B. 3%.

C. 4%.

Solution

The correct answer is B.

First, you should re-arrange the data in ascending order:

{2% 2% 2.09% 2.7% 3% 3% 4% 7% 7.2% 7.5% 8%}

Secondly, you should establish the 5th decile. This is simply the 50th percentile and is

actually the median:

(1 + 11) 50
P 50 =
100
= 12 × 0.5
= 6 , i.e., the 6th data point

Therefore, the 5th decile = 50th percentile = Median = 3%

86
© 2014-2024 AnalystPrep.
LOS 3b: calculate, interpret, and evaluate measures of dispersion to
address an investment problem

Measures of dispersion are used to describe the variability or spread in a sample or population.

They are usually used in conjunction with measures of central tendency, such as the mean and

the median. Specifically, measures of dispersion are the range, variance, absolute deviation, and

standard deviation.

Measures of dispersion are essential because they give us an idea of how well the measures of

central tendency represent the data. For example, if the standard deviation is large, then there

are large differences between individual data points. Consequently, the mean may not be

representative of the data.

Range

The range is the difference between the highest and the lowest values in a dataset, i.e.,

Range = Maximum value - Minimum value

Example: Calculating the Range

Consider the following scores of 10 level I candidates:

{78 56 67 51 43 89 57 67 78 50}

Range = 89 − 43 = 46

Advantage of the Range

The range is easy to compute.

Disadvantages of the Range

The range is not a reliable dispersion measure. It provides limited information about

the distribution because it uses only two data points.

87
© 2014-2024 AnalystPrep.
 The range is sensitive to outliers.

Mean Absolute Deviation (MAD)

MAD is a measure of dispersion representing the average of the absolute values of the

deviations of individual observations from the arithmetic mean. Therefore,

∑ ∣∣Xi − X̄∣∣
MAD =
n

Remember that the sum of deviations from the arithmetic means is always zero, which is why we

use absolute values.

Example: Calculating Mean Absolute Deviation

Six financial analysts have reported the following returns on six different large-cap stocks over

2021:

{6% 7% 12% 2% 3% 11%}

Calculate the mean absolute deviation and interpret it.

Solution

First, we have to calculate the arithmetic mean:

(6% + 7% + 12% + 2% + 3% + 11%)

X̄ = = 6.83%
6

Next, we can now compute the MAD:

{|6%– 6.83%| + |7%– 6.83%| + |12%– 6.83%| + |2%– 6.83%| + |3%– 6.83%| + |11%– 6.83
MAD =
6
0.83 + 0.17 + 5.17 + 4.83 + 3.83 + 4.17
=
6
= 3.17%

88
© 2014-2024 AnalystPrep.
Interpretation: On average, an individual return deviates by 3.17% from the mean return of

6.83%.

Sample Variance and Sample Standard Deviation

The sample variance,s2 , is the measure of dispersion that applies when working with a sample

instead of a population.

2
2
∑ (Xi − X̄)
s =
n −1

Where:

X̄ = Sample mean.

n = Number of observations.

Note that we are dividing by n– 1. This is necessary to remove bias.

The sample standard deviation, s, is simply the square root of the sample variance.

2
(Xi − X̄ )
s = √s = ⎷
2
n− 1

Example: Calculating Sample Mean and Variance

Assume that the returns realized in the previous example were sampled from a population

comprising 100 returns. The sample mean and the corresponding sample variance are closest to:

Solution

The sample mean will still be 6.83%.

Hence,

89
© 2014-2024 AnalystPrep.
 {(6% − 6.83%)2 + (7% − 6.83%)2 + (12% − 6.83%)2 + (2% − 6.83%)2 + (3% − 6.83%)2 + (11%
2
s =
5
= 0.001656

Therefore,

1
s = 0.001656 2
= 0.0407

Downside Deviation and Coefficient of Variation

When trying to estimate downside risk (i.e., returns below the mean), we can use the

following measures:

Semi-variance: The average squared deviation below the mean.

Semi-deviation (also known as semi-standard deviation): The positive square root of

semi-variance.

Target semi-variance: The sum of the squared deviations from a specific target

return.

Target semi-deviation: The square root of target semi-variance.

Sample Target Semi-Deviation

The target semi deviation, sTarget , is calculated as follows:

n (Xi − B)2
sTarget = ∑
⎷for all X n−1
i ≤B

Where B is the target and n is the total number of sample observations.

Yearly returns of an equity mutual fund are provided as follows.

90
© 2014-2024 AnalystPrep.
 Month Return %
2010 36%
2011 29%
2012 10%
2013 52%
2014 41%
2015 16%
2016 10%
2017 23%
2018 −10%
2019 −19%
2020 2%

What is the target downside deviation if the target return is 20%?

Solution

Squared
Deviation Deviation
Return deviations
Month from the 20% below the
% below the
target target
target
2010 36.00 16.00 − −
2011 29.00 9.00 − −
2012 10.00 (10.00) (10.00) 100
2013 52.00 32.00 −
2014 41.00 21.00 −
2015 16.00 (4.00) (4.00) 16
2016 10.00 (10.00) (10.00) 100
2017 23.00 3.00 −
2018 (10.00) (30.00) (30.00) 900
2019 (19.00) (39.00) (39.00) 1, 521
2020 2.00 (18.00) (18.00) 324
Sum 2,961

Here n = 11– 1 = 10 so that:

0.5
2961
Target semi-deviation = ( ) = 17.21%
10

Coefficient of Variation

91
© 2014-2024 AnalystPrep.
The coefficient of variation, CV , is a measure of spread that describes the amount of variability

of data relative to its mean. It has no units, so we can use it as an alternative to the standard

deviation to compare the variability of data sets that have different means. The coefficient of

variation is given by:

s
CV =
X̄

Where:

s = Standard deviation of a sample.

X̄ = Mean of the sample.

?
Note: The formula can be replaced with ?
when dealing with a population.

Procedure to Follow While Calculating the Coefficient of Variation:

1. Compute the mean of the data.

2. Calculate the sample standard deviation of the data set, s.

3. Find the ratio of s to the mean, x? .

Example: Coefficient of Variation

What is the relative variability for the samples 40, 46, 34, 35, and 45 of a population?

Solution

Step 1: Calculate the mean.

(40 + 46 + 34 + 35 + 45) 200

Mean = = = 40
5 5

Step 2: Calculate the sample standard deviation. (Start with the variance, s2.)

(40– 40)2 + . .. +(45– 40)2

92
© 2014-2024 AnalystPrep.
 (40– 40)2 + . .. +(45– 40)2
s2 =
4
122
=
4
= 30.5

Note: Since it is the sample standard deviation (not the population standard deviation), we use

n − 1 as the denominator.

Therefore,

s = √30.5 = 5.52268

Step 3: Calculate the ratio.

Mean 5.52268
= = 0.13806 or 13.81%
s 40

Interpreting the Coefficient of Variation

In finance, the coefficient of variation is used to measure the risk per unit of return. For

example, imagine that the mean monthly return on a T-Bill is 0.5% with a standard deviation of

0.58%. Suppose we have another investment, say, Y, with a 1.5% mean monthly return and

standard deviation of 6%; then,

0.58
CVT−Bill = = 1.16
0.5

6
CVY = =4
1.5

Interpretation: The dispersion per unit monthly return of T-Bills is less than that of Y.

Therefore, investment Y is riskier than an investment in T-Bills.

93
© 2014-2024 AnalystPrep.
Question 1

If a security has a mean expected return of 10% and a standard deviation of 5%, its

coefficient of variation is closest to:

A. 0.005.

B. 0.500.

C. 2.000.

Solution

The correct answer is B.

S 0.05
CV = = = 0.5
x? 0.10

Where:

s = The standard deviation of the sample.

x? = The mean of the sample.

A is incorrect. It assumes the following calculation.

0.05
CV = = 0.005
10

C is incorrect. It assumes the following calculation.

10
CV = =2
5

Question 2

You have been given the following data:

94
© 2014-2024 AnalystPrep.
{12 13 54 56 25}

Assuming that this is a sample from a certain population, the sample standard

deviation is closest to:

A. 21.62.

B. 374.00.

C. 1,870.00.

The correct answer is A.

(12 + 13 + ⋯ + 25) 160

X̄ = = = 32
5 5

Hence,

{(12 − 32)2 + (13 − 32)2 + (54 − 32)2 + (56 − 32)2 + (25 − 32)2 }
s2 =
4
1870
= = 468
4

Therefore,

s = √468 = 21.62

95
© 2014-2024 AnalystPrep.
LOS 3c: interpret and evaluate measures of skewness and kurtosis to
address an investment problem

Since the deviations from the mean are squared when calculating variance, we cannot determine

whether significant deviations are more likely to be positive or negative. In order to identify

other crucial distributional traits, we must look beyond measures of central tendency, location,

and dispersion.

Skewness

Skewness refers to the degree of deviation from a symmetrical distribution, such as the normal

distribution. A symmetrical distribution has identical shapes on either side of the mean.

Distributions that are nonsymmetrical have unequal shapes on either side of the mean, leading to

skewness. This is because nonsymmetrical distributions depart from the usual bell shape of the

96
© 2014-2024 AnalystPrep.
normal distribution.

Skewness can be positive, negative, or, in some cases, undefined. The shape of a skewed

distribution depends on outliers, which are extremely negative and positive observations.

Positive Skewness

A positively skewed distribution has a long right tail because of many outliers or extreme

values on the right side. Perhaps the best way to remember its shape is to consider its points in a

positive direction. Most data points are concentrated on the left side.

An example of a positively skewed distribution would be the income of individuals living in a

specific country.

Negative Skewness

A negatively skewed distribution has a long left tail resulting from many outliers on the left side

of the distribution. Therefore, we could say that it points in the negative direction. This is

97
© 2014-2024 AnalystPrep.
because the right side harbors most of the data points.

Application of Skewness

Skewness matters in finance. Market data often show positive or negative skewness, like stock

prices or mortgage costs. Investors can predict if future prices will be above or below the mean

based on the skewness of the market segment.

Calculating Sample Skewness

The approximate sample skewness when sample is large (n ≥ 100) is given by:

3
1 ∑n (Xi − X̄)
Skewness = ( ) i=1
n s3

Where:

X̄ = Sample mean.

98
© 2014-2024 AnalystPrep.
s = Sample standard deviation.

n = Number of observations.

A positive value indicates positive skewness. A 'zero' value indicates that the data is not skewed.

Lastly, a negative value indicates negative skewness or a negatively skewed distribution.

Example: Calculating Skewness

Suppose we have the following observations:

{12 13 54 56 25}

What is the skewness of the data?

Solution

First, we must determine the sample mean and the sample standard deviation:

(12 + 13 + 54 + 56 + 25) 160

X̄ = = = 32
5 5
(12 − 32)2 + (13 − 32)2 + … (25 − 32)2
s2 =
4
= 467.5

Therefore,

s = √467.5 = 21.62

Now we can work out the skewness:

3
1 ∑ni 1 (X i − X̄)
Skewness = ( ) =
n s3
3 3 3
1 (−20) + (−19) + 22 3 + 243 + (−7)
Skewness = ( )
5 21.62 3
Skewness = 0.1835

Skewness is positive. Hence, the data has a positively skewed distribution.

99
© 2014-2024 AnalystPrep.
Kurtosis

Kurtosis refers to the measurement of the degree to which a given distribution is more or less

'peaked' relative to the normal distribution. The concept of kurtosis is instrumental in decision-

making. In this regard, we have three categories of distributions:

Leptokurtic.

Mesokurtic.

Platykurtic.

Leptokurtic

A leptokurtic distribution is more peaked than the normal distribution. The higher peak results

from the clustering of data points along the x-axis. The tails are also fatter than those of a normal

distribution. The coefficient of kurtosis is usually more than 3.

The term "lepto" means thin or skinny. When analyzing historical returns, a leptokurtic

distribution means that small changes are less frequent since historical values are clustered

100
© 2014-2024 AnalystPrep.
around the mean. However, there are also large fluctuations represented by the fat tails.

Platykurtic

A platykurtic distribution has extremely dispersed points along the x-axis, resulting in a lower

peak when compared to a normal distribution. "Platy" means broad. Hence, the prefix fits the

distribution's shape, which is wide and flat. The points are less clustered around the mean

compared to a leptokurtic distribution. The coefficient of kurtosis is usually less than 3.

Returns that follow this type of distribution have fewer major fluctuations compared to

leptokurtic returns. However, you should note that fluctuations represent the riskiness of an

asset. More fluctuations represent more risk and vice versa. Therefore, platykurtic returns are

less risky than leptokurtic returns.

Mesokurtic

Lastly, mesokurtic distributions have a curve that is similar to that of a normal distribution. In

other words, the distribution is mainly normal.

The majority of equity return series are found to have fat tails. Suppose a return distribution has

fat tails, and we apply statistical models that do not consider distribution. In that case, we will

overestimate the probability of either extremely poor or very favorable outcomes.

Investors often study a stock's daily trading volume distribution to assess its trading liquidity. It

helps them see if the market can handle a large trade in that stock. This is useful for investors

who want to make big investments or exit their positions in a particular stock.

Calculating Sample Kurtosis

Sample kurtosis is always measured relative to the kurtosis of a normal distribution, which is 3.

Therefore, we are always interested in the "excess "kurtosis, i.e.,

Excess kurtosis = Sample kurtosis– 3

Where:
4
¯
101
© 2014-2024 AnalystPrep.
 n 4
1 ∑ (Xi − X̄)
Sample Excess Kurtosis = ( ) i =1
n s4

Positive excess kurtosis indicates a leptokurtic distribution. A zero value indicates a mesokurtic

distribution. Lastly, a negative excess kurtosis represents a platykurtic distribution.

Example: Calculating Kurtosis

Using the data from the example above (12, 13, 54, 56, and 25), determine the type of kurtosis

present.

(12 + 13 + 54 + 56 + 25) 160

X̄ = = = 32
5 5
(12 − 32)2 + (13 − 32)2 + … (25 − 32)2
s2 = = 467.5
4
s = √467.5 = 21.62

Therefore,

n 4
1 ∑i 1 (Xi − X̄ )
Excess Kurtosis = ( ) = −3
n s4
4 4 4 4 4
1 (−20) + (−19) + 22 + 24 + (−7)
Excess Kurtosis = ( ) −3
5 21.624
Excess Kurtosis = 2.2139

Since the excess kurtosis is negative, we have a platykurtic distribution.

102
© 2014-2024 AnalystPrep.
Question 1

The skewness of the normal distribution is most likely:

A. Zero.

B. Positive.

C. Negative.

Solution

The correct answer is A.

Since the normal curve is symmetric about its mean, its skewness is zero.

B is incorrect because a positively skewed distribution has positive skewness.

C is incorrect because a negatively skewed distribution has negative skewness.

Question 2

A frequency distribution in which there are too few scores at the extremes of the

distribution is most likely called:

A. Platykurtic.

B. Leptokurtic.

C. Mesokurtic.

Solution

The correct answer is A.

A platykurtic distribution has "thin" tails and is flatter compared to a normal

distribution. It implies that there are fewer scores at the extremes of the distribution,

which aligns with the question's description.

103
© 2014-2024 AnalystPrep.
Question 3

When most of the data are concentrated on the left of the distribution, it is most likely

called:

A. Symmetric distribution.

B. Positively skewed distribution.

C. Negatively skewed distribution.

Solution

The correct answer is B.

A distribution is said to be skewed to the right or positively skewed when most of the

data are concentrated on the left of the distribution. A distribution is said to be

skewed to the left or negatively skewed if most of the data are concentrated on the

right of the distribution. The left tail clearly extends farther from the distribution's

center than the right tail.

104
© 2014-2024 AnalystPrep.
A is incorrect. A symmetric distribution is one in which the left and right sides

mirror each other.

C is incorrect. A distribution is said to be skewed to the left or negatively skewed if

most of the data are concentrated on the right of the distribution. The left tail extends

farther away from the mean than the right tail.

105
© 2014-2024 AnalystPrep.
LOS 3d: Interpret the correlation between two variables to address an
investment problem

Covariance

Covariance is a measure of how two variables move together. The sample

covariance of X and Y is calculated as follows:

∑N ¯
i=1 (Xi − X̄) (Yi − Y )
sX Y =
n −1

The formula above implies that the sample covariance is the mean of the product of the

deviations in the two random variables and from their sample means.

If the covariance between two random variables is positive, it means they move in the same

direction. When one is below its mean, the other is also below its mean, and vice versa.

A major drawback of covariance is that it is difficult to interpret since its value can vary from

negative infinity to positive infinity.

Correlation

Correlation is a standardized measure of the linear relationship between two variables. It takes

the covariance and divides it by the product of the standard deviations of both variables. As a

result, its value ranges between -1 and +1 and is easier to interpret.

The sample correlation coefficient is calculated as follows:

Sxy
r xy =
Sx × Sy

Where:

sXY = Covariance between variable X and Y.

106
© 2014-2024 AnalystPrep.
sX = Standard deviation of variable X.

sY = Standard deviation of variable Y.

Properties of Correlation

Correlation ranges between -1 to +1 for two random variables, X and Y.

A correlation of 0 (uncorrelated variables) indicates no linear (straight line)

relationship exists between the variables.

107
© 2014-2024 AnalystPrep.
 A positive correlation close to +1 indicates a strong positive linear relationship.

A correlation of 1 indicates a perfect linear relationship.

A negative correlation close to -1 indicates a strong negative linear relationship.

A correlation of -1 indicates a perfect inverse linear relationship.

Limitations of Correlation Analysis

Two variables can have a very low correlation despite having a strong nonlinear

relationship.

Correlation can be an unreliable measure when outliers are present in the data.

Correlation does not imply causation. This implies that the correlation may be spurious.

A spurious correlation refers to:

Correlation between two variables due to chance relationships in a particular

dataset.

Correlation arising between variables when they are divided by a third

variable.

Correlation between two variables arising from their relation to a third

variable.

108
© 2014-2024 AnalystPrep.
Question

The correlation coefficient between X and Y is 0.7, and the covariance is 29. If the

variance of Y is 25, the variance of X is closest to:

A. 8.29.

B. 29.

C. 68.65.

Solution

The correct answer is C.

sX Y
rX Y =
sX × SY
29
⇒ 0.7 =
X ×5
∴ X = 8.2857

Variance = 8.28572 = 68.65

109
© 2014-2024 AnalystPrep.
 Learning Module 4: Probability Trees and Conditional Expectations

LOS 4a: calculate expected values, variances, and standard deviations

and demonstrate their application to investment problems

Expected Value

Expected value is an essential quantitative concept investors use to estimate investment returns

and analyze any factor that may impact their financial position.

Mathematically, the expected value is the probability-weighted average of the possible outcomes

of the random variable. For a random variable X, the expected value of X is denoted E(X). More

specifically,

E(X) = P (X1 )X1 + P (X2 )X2 + ⋯ + P (Xn )Xn

n
= ∑ P (Xi )X i
i =1

Where,

Xi = One of n possible outcomes of the discrete random variable X.

P (Xi ) = P (Xi = xi ) = Probability of X taking the value x.

Note that the expected can be a forecast (looking into the future) or the true value of the

population mean.

The sample mean differs from the expected value. The sample mean is a central value for a

specific set of observations, calculated as an equally weighted average of those observations.

Example: Calculating Expected Value

An analyst anticipates the following returns from an asset:

Return Probability
5% 65%
7% 25%
8% 10%

110
© 2014-2024 AnalystPrep.
The expected value of the investment is closest to:

Solution

Recall that,

n
E(X) = ∑ P(Xi )Xi
i=1

= 0.05 × 0.65 + 0.07 × 0.25 + 0.10 × 0.08

= 0.0325 + 0.0175 + 0.008
= 0.058 = 5.8%

Variance and Standard Deviation

Consider expected value as a forecast of the outcome of an investment. Then, variance and

standard deviation measure the risk of an investment. That is the dispersion of outcomes around

the mean.

The variance of a random variable is the expected value (the probability-weighted average) of

squared deviations from the random variable's expected value. Denoted by σ 2 (X) or V ar(X), its

formula is given by:

V ar(X) = E[X − E(X)]2

2 2
= P (X1 )[X1 − E(X)] + P (X2 )[X2 − E(X)] + ⋯
+ P (Xn )[Xn − E(X)]2
n
⇒ V ar(X) = ∑ P (Xi )[Xi − E(X)]2
i =1

Since variance is in squared terms, it can take any number greater than or equal to

0(V ar(X) ≥ 0) . Intuitively, if V ar(X) = 0, there is no risk (dispersion). On the other hand, if

V ar(X) > 0, it signifies the dispersion of outcomes.

Moreover, V ar(X) is a quantity given in square units of X. That is, if the X is given in percentage,

then V ar(X) is given in squared percentage.

The standard deviation is the square root of variance:

111
© 2014-2024 AnalystPrep.
 σ(X) = √σ 2(X) = √V ar(X)

The standard deviation is given in the same units as the random variance; hence it is easy to

interpret.

112
© 2014-2024 AnalystPrep.
Question

An analyst anticipates the following returns from an asset:

Return Probability
5% 65%
7% 25%
8% 10%

The variance and standard deviation of the investment are closest to:

Solution

We know that,

n
V ar(X) = ∑ P(Xi )[Xi − E(X)]2
i=1

= 0.65(0.05 − 0.058)2 + 0.25(0.07 − 0.058)2

+ 0.10(0.08 − 0.058)2
= 0.000126

For standard deviation,

σ(X) = √V ar(X) = √0.000126 = 0.0112

113
© 2014-2024 AnalystPrep.
LOS 4b: Formulate an investment problem as a probability tree and
explain the use of conditional expectations in investment application

A tree diagram is a visual representation of all possible future outcomes and the associated

probabilities of a random variable. Tree diagrams are handy when we have several possible

outcomes.

They facilitate the recording of all the possibilities in a clear, uncomplicated manner. Each

branch in a tree diagram represents an outcome.

Example: Probability Tree

Let's consider a scenario where we toss a fair coin twice. The outcomes of these tosses are

independent, meaning the first toss doesn't influence the second one.

For the first toss, we have two possibilities: It can result in either a head or a tail. Similarly, we

still have the same two possibilities for the second toss: head or tail. Importantly, the outcome of

the second toss is not affected by what happened in the first toss because coins don't have

memory.

We can visualize these probabilities using a tree diagram like this:

114
© 2014-2024 AnalystPrep.
Please, note the following:

The tree diagram must include all the possible outcomes.

The sum of the probabilities must add up to 1.

The number of branches is the number of different possibilities.

The numbers on the branches present probabilities.

To calculate probabilities, we follow the tree branches from left to right and multiply any

probabilities we encounter.

So, to find the probability of getting two heads (HH), we multiply the probabilities along the

path.

1 1 1
115
© 2014-2024 AnalystPrep.
 1 1 1
P(HH ) = × =
2 2 4

If we sum up the probabilities of all possibilities, we get 1.

Conditional Expectations

In investments, conditional expectation means predicting the expected value of an investment

based on specific real-world events. Analysts consider the probability and impact of future events

when calculating this value.

Competitors, governments, and other financial institutions keep releasing new information. Such

pieces of information may have a positive or a negative impact on investment. This means that a

project's expected value must be based on real-world dynamics.

In statistics, the conditional expected value is the expected value of a random variable X given an

event or scenario S , denoted by E(X ∣ S ).

Now, assume that X can take on any of n different outcomes X1 , X2 , … , Xn, which are outcomes

from a set of mutually exclusive and exhaustive events. Then,

E(X ∣ S) = X1 ⋅ P (X1 ∣ S) + X2 ⋅ P (X 2 ∣ S) + ⋯ + Xn ⋅ P(Xn ∣ S)

n
= ∑ Xi ⋅ P (Xi ∣ S)
i=1

Stating Unconditional Expected Value in Terms of Conditional

Expected Value

To state the unconditional expected values in terms of conditional expected value, we use the

total probability rule for expected value, which is built from the following formula:

E(X) = P (S) ⋅ E(X ∣ S) + P (SC ) ⋅ E(X ∣ S C )

Where S C is the complement (event or scenario “S” does not occur) of S .

116
© 2014-2024 AnalystPrep.
Now assume that S can take on any of S1 , S2 , … , Sn mutually exclusive and exhaustive scenarios

or events, then.

E(X) = P (S1 ) ⋅ E(X ∣ S1 ) + P (S2 ) ⋅ E(X ∣ S2 ) + ⋯ + P(Sn ) ⋅ E(X ∣ Sn )

n
= ∑ P(Si ) ⋅ E(X ∣ Si)
i=1

Example: Conditional Expectation

The probability of relaxed trade restrictions in a given country is 40%. Therefore, shareholders of

XYZ Company Limited expect a 5% share return if trade restrictions are maintained and a loss of

8% if they are relaxed. The expected change in return is closest to:

Solution

We must take every possibility into account. We have a 40% chance of relaxed trade restrictions

in this case. Intuitively, this means there is a 60% chance that the current restrictions will be

maintained. Therefore:

n
E(X) = ∑ P (Si) ⋅ E(X ∣ Si )
i =1

= 0.6(0.05) × 0.4(−0.08)
= −0.002

Example: Total Probability Rule for Expected Value

BlueChip Inc.'s profits are sensitive to economic growth, benefitting significantly during periods

of high economic growth. Suppose there is a 0.70 probability that BlueChip Inc. will operate in a

high-growth economic environment in the next fiscal year and a 0.30 probability that it will

operate in a moderate-growth environment. Assume that the chance of a recession is negligible.

If a high-growth economic environment occurs, the probability that EPS will be USD 3.00 is

estimated at 0.40, and the probability that EPS will be USD 2.80 is estimated at 0.60.

On the other hand, if the company operates in a moderate-growth environment, the probabilities

that the EPS will be USD 2.50 and USD 2.25 are 20% and 80%, respectively.

117
© 2014-2024 AnalystPrep.
Calculate the expected value of EPS for BlueChip Inc. in the next fiscal year.

Solution

We start by drawing a probability tree:

We first need to calculate the conditional expectations of EPS for each scenario: High-growth

environment and moderate-growth environment. That is,

E(EPS ∣ High-growth environment) = 0.40 × 3.00 + 0.60 × 2.80

= USD 2.88
E(EP S ∣ Moderate-growth environment) = 0.20 × 2.50 + 0.80 × 2.25
= USD 2.30

Using the total probability for the expected value, we have:

118
© 2014-2024 AnalystPrep.
E(EP S) = P (High-growth environment)
⋅ E(EP S ∣ High-growth environment)
+ P (High-growth environment)
⋅ E(EP S?Moderate-growth environment)
= 0.70 × 2.88 + 0.30 × 2.30
= USD 2.71

119
© 2014-2024 AnalystPrep.
Question 1

Someone picks a card from an ordinary pack of 52 playing cards without

replacement. He then picks another one. Draw a probability tree and use it to

calculate the probability of picking two red cards.

25
A. 102
.
13
B. 51
.
26
C. 51
.

Solution

The correct answer is A.

120
© 2014-2024 AnalystPrep.
Question 2

There is a 20% chance that the government will impose a tariff on imported cars. A

company that assembles cars locally expects returns of 14% if the tariff is imposed

and returns of 11% otherwise. The (unconditional) expected return is closest to:

A. 11.6%.

B. 12.8%.

C. 12.5%.

Solution

The correct answer is A.

The unconditional expected return will be the sum of:

1. The expected return given no tariff times the probability that a tariff

will not be imposed.

2. The expected return given tariff times the probability that the tariff

will be imposed. Therefore,

n
E(X) = ∑ P (Si) + E(X ∣ Si )
i =1

= 0.11(0.8) + 0.14(0.2)
= 0.116 = 11.6%

121
© 2014-2024 AnalystPrep.
LOS 4c: calculate and interpret an updated probability in an investment
setting using Bayes’ formula

Investors make investment decisions based on their experience and expertise. Their decisions

may change in the wake of new knowledge and observations.

Bayes' formula allows us to update our decisions as we receive new information. In other words,

Bayes' formula is used to calculate an updated or posterior probability given a set of prior

probabilities for a given event.

Given a set of prior probabilities for an event, if we receive new information, the updated

probability is as follows:

Probability of the new information given event

Updated probability of an event given the new information=
Unconditional probability of the new information
× Prior probability of event.

The above equation can be written as:

P(Information ∣ Event)
P (Event ∣ Information) = ⋅ P (Event)
P (Information)

Deriving Bayes' Formula

Let B1 , B2 , B3 , … , Bn be a set of mutually exclusive and exhaustive events.

Using the conditional probability:

P (Bi ∩ A)
P (Bi ∣ A) = … … (1)
P (A)

And also, the relationship:

P (Bi ∩ A) = P (A ∩ Bi ) = P(B i) ⋅ P (A ∣ B i) … … (2)

Also, using the total probability rule:

122
© 2014-2024 AnalystPrep.
 n n
P (A) = ∑ P (A ∩ B i) = ∑ P (B i) ⋅ P (A ∣ Bi ) … … (3)
i=1 i=1

Substituting equations (2) and (3) in (1), we have:

(P (A ∣ B i)
P (Bi ∣ A) = ⋅ P (Bi )
∑ni=1 P (Bi ) ⋅ P(A ∣ Bi )

This is the Bayes' formula, and it allows us to ‘turnaround’ conditional probabilities, i.e., we can

calculate P (Bi ∣ A) if given information only about P (A ∣ B i ).

Note that:

1. P (Bi ) are known as prior probabilities.

2. Event A is some event known to have occurred.

3. P (Bi ∣ A) is the posterior probability.

Example: Bayes' Formula

An Investment Analyst wishes to investigate the performance of stocks by considering a number

of stocks listed on different exchanges. In the sample, 50% of stocks were listed on the New York

Stock Exchange (NYSE), 30% on the London Stock Exchange (LSE), and 20% on the Tokyo Stock

Exchange (TSE).

The probability of a stock posting a negative return on the NYSE, LSE, and TSE is 40%, 35%, and

25%, respectively.

If the Analyst picks a stock at random from this group, what is the probability that it has a

negative return on the NYSE?

Solution

We are looking for P(NYSE | Negative Return).

Let’s define the following events:

NYSE is the event “A stock chosen at random is listed on the NYSE.”

123
© 2014-2024 AnalystPrep.
LSE is the event “A stock chosen at random is listed on the LSE.”

TSE is the event “A stock chosen at random is listed on the TSE.”

Finally, let NR be the event “A randomly chosen stock posts a negative return.”

Therefore,

P (N Y SE) P(N R|NY SE)

P (N Y SE|N R) =
P (N Y S E) P (N R|NY SE) + P (LSE) P (N R|LSE) + P (T SE) P (N R|TS E)
0.5 × 0.4
=
0.5 × 0.4 + 0.3 × 0.35 + 0.2 × 0.25
0.2
=
0.355
= 0.5634 ≈ 56.3%

124
© 2014-2024 AnalystPrep.
Question

You have developed a set of criteria for assessing potential investments in growth-

stage companies. Companies not meeting these criteria are predicted to be insolvent

within 24 months. You gathered the following information when validating your

criteria:

Fifty percent of the companies that have been assessed will become

insolvent within 24 months: P (insolvency) = 0.50.

Sixty-five percent of the companies assessed meet the criteria:

P (meet criteria) = 0.65 .

The probability that a company will meet the criteria given that it remains

solvent for 24 months is 0.80: P (meet criteria ∣ solvency) = 0.80 .

The probability that a company will remain solvent, given that it meets the criteria,

that is, P (solvency ∣ meet criteria), is closest to:

A. 20%.

B. 50%.

C. 62%.

Solution

Using Bayes' formula, we have:

P(solvency ∣ meet criteria)

P (meet criteria ∣ solvency)P (solvency)
=
[P (meet criteria ∣ solvency)P (solvency)
+P (meet criteria ∣ insolvency)P (insolvency)]
0.80 × 0.50
=
0.80 × 0.50 + P (meet criteria ∣ insolvency) × 0.50

Clearly, we need to calculate the P (meet criteria ∣ insolvency). Using the total

probability:

125
© 2014-2024 AnalystPrep.
 P (meet criteria) = P (meet criteria ∣ solvency)P (solvency)
+ P (meet criteria ∣ insolvency)P (insolvency)
⇒ 0.65 = 0.80 × 0.50 + P(meet criteria ∣ insolvency) × 0.50

0.65 − 0.80 × 0.50

∴ P (meet criteria ∣ insolvency) = = 0.50
0.50

As such,

0.80 × 0.50
P (solvency ∣ meet criteria) = = 0.6153 ≈ 62%
0.80 × 0.50 + 0.50 × 0.50

126
© 2014-2024 AnalystPrep.
 Learning Module 5: Portfolio Mathematics

LOS 5a: calculate and interpret the expected value, variance, standard
deviation, covariances, and correlations of portfolio returns

A portfolio is a collection of investments a company, mutual fund, or individual investor holds. A

portfolio consists of assets such as stocks, bonds, or cash equivalents. Financial professionals

usually manage a portfolio.

Portfolio Expected Return

To calculate the portfolio's expected return, you take the expected returns of each security in the

portfolio. Then, you multiply each security's expected return by its proportion in the portfolio and

add them up. The formula below helps you find the portfolio's expected return:

E(Rp ) = w 1 E(R 1 ) + w 2 E(R 2 ) + ⋯ w n E(R n )

Where:

w 1 , w 2 , … , w n = Weights (market value of asset/market value of the portfolio) attached to assets

1, 2 ,… , n.

R 1, R 2 , … , R n = Expected returns for assets 1 , 2, … , n .

Example: Portfolio Expected Return

Assume we have a simple portfolio of two mutual funds, one invested in bonds and the other

invested in stocks. Let us further assume that we expect a stock return of 8% and a bond return

of 6%, and our allocation is equal in both funds. The expected return would be calculated as

follows:

E(Rp) = (0.5 × 0.08) + (0.5 × 0.06) = 0.07 or 7%

Portfolio Variance

The variance of a portfolio's return is a function of the individual asset covariances as well as the

127
© 2014-2024 AnalystPrep.
covariance between each of them.

Consider a portfolio with three assets: A, B, and C. The portfolio variance is given by:

Portfolio Variance
= WA2 σ 2(R A) + WB2 σ 2(R B) + WC2 σ 2(R C ) + 2(WA )(W B)Cov(RA , RB )
+ 2(WA )(WC )Cov(RA , R C ) + 2(WB )(WC )Cov(RB , RC )

If we have two assets, A and B, then:

Portfolio Variance = WA2 σ 2(R A) + WB2 σ 2(R B) + 2(W A)(WB )Cov(R A, R B)

Where:

WA = Weight of assets A in the portfolio.

WB = Weight of assets B in the portfolio.

σ 2(R A) = Variance of the returns on assets A.

σ 2(R B) = Variance of the returns on assets B .

Portfolio variance is a measure of risk. The higher the variance, the higher the risk. Investors

usually reduce the portfolio variance by choosing assets with low or negative covariance, e.g.,

stocks and bonds.

Portfolio Standard Deviation

Portfolio standard deviation is simply the square root of the portfolio variance. It is a measure of

the riskiness of a portfolio.

Considering a portfolio with two assets, A and B, the portfolio standard deviation is given by:

Standard deviation = √(W 2 σ 2 (R A ) + WB2 σ 2 (R B ) + 2(WA)(WB )Cov(RA ,R B )

A

Covariance

128
© 2014-2024 AnalystPrep.
Covariance is a measure of the degree of co-movement between two random variables. The

general formula used to calculate the covariance between two random variables, X and Y is:

Cov(X, Y ) = σ(X, Y ) = E[(X − E[X])(Y − E[Y ])

Where:

Cov(X, Y ) = Covariance of X and Y .

E[X] = Expected value of the random variable X.

E[Y ] = Expected values of the random variable Y.

This formula calculates the population covariance. It does this by taking the probability-weighted

average of the cross-products of the random variables' deviations from their expected values for

every possible outcome.

Sample Covariance

The sample covariance between two variables, X and Y , based on a sample data of size n is:

n (Xi − X̄ )(Y i − Y¯ )
Cov(X, Y ) = ∑
i=1 n− 1

Where:

X̄ = Sample mean of X .

Y¯ = Sample mean of Y .

Xi and Y i = i-th data points of X and Y , respectively.

The covariance between two random variables can be positive, negative, or zero.

A positive number indicates co-movement. The variables tend to move in the same

direction.

A value of zero indicates no relationship.

129
© 2014-2024 AnalystPrep.
 A negative value shows that the variables move in opposite directions.

Covariance Matrix

A covariance matrix displays a complete list of covariances between assets needed to calculate

the portfolio variance. Consider a portfolio with three assets A, B, and C. The covariance matrix

is as follows:

Asset A B C
A Cov(RA , RA ) Cov(RA , RB ) Cov(RA , R C )
B Cov(RB , RA ) Cov(RB , RB ) Cov(RB , RC )
C Cov(RC , R A) Cov(RC , R B) Cov(R C , RC )

The off-diagonal (bolded) terms represent variances since, for example:

Cov(R A, R A) = ρ(A, A)σAσA = 1?σA2 = σA2

As such, the table above transforms:

Asset A B C
A σA2 Cov(R A, R B) Cov(RA ,R C )
B Cov(R B, R A) σB2 Cov(RB , RC )
C C ov(R C , RA ) Co v(R C , RB ) σC2

Intuitively, a three-asset portfolio would have 3 × 3 = 9 entries of covariances. However, we do

not count the off-diagonal terms since they contain the individual variances of the assets. As

such, we have 6(= 9 − 3) covariances.

Note that:

C ov(R B, R A) = Cov(RA ,R B )
Cov(RA ,R C ) = Cov(RA ,R C )
Cov(R C , RB ) = Cov(RB , RC )

6
Therefore, there are = 3 distinct covariance terms in the above covariance matrix.
2

n( n −1)
In general, if we have n securities in a portfolio, there are distinct covariances and n
2

130
© 2014-2024 AnalystPrep.
variances to estimate.

Correlation

Correlation is the covariance ratio between two random variables and the product of their two

standard deviations. The correlation formula for random variables X and Y is:

Correlation (X,Y) = Co rr(X, Y ) = ρ(X, Y )

Cov(X,Y )
=
Standard deviation(X) × Standard deviation(Y)
Cov(X,Y )
=
σX σY

Correlation measures the strength of the linear relationship between two variables. While the

covariance can take on any value between negative infinity and positive infinity, the correlation is

always between -1 and +1:

+1 indicates a perfect linear relationship (i.e., the two variables move in the same

direction with equal unit changes).

Zero indicates no linear relationship at all.

-1 indicates a perfect inverse relationship, i.e., a unit change in one means that the

other will have a unit change in the opposite direction.

Example: Calculating Correlation Coefficient from the Covariance Matrix #1

Harrison is a portfolio manager who oversees three assets: A, B, and C. The covariance matrix of

these assets is shown below:

Asset A B C
A 0.04 0.02 0.01
B 0.02 0.05 0.015
C 0.01 0.015 0.09

Using this information, what is the correlation coefficient between assets B and C ?

Solution

131
© 2014-2024 AnalystPrep.
Note:

Cov(B , C)
Correlation (B, C) =
σBσC
0.015
= = 0.224
√0.05 × 0.09

Example: Calculating the Correlation Coefficient #2

We expect a 15% chance that ABC Corp's stock returns for the next year will be 6%. There's a

60% probability that they will be 8% and a 25% probability of a 10% return. The expected return

is 8.2%, and the standard deviation is 1.249%.

We also anticipate that the same probabilities and states are associated with a 4%, 5%, and 5.5%

return for XYZ Corp. The expected value of returns is then 4.975%, and the standard deviation is

0.46%.

To calculate the covariance and the correlation between ABC and XYZ returns, then:

Cov(RABC , R XY Z ) = 0.15(0.06 − 0.082)(0.04 − 0.04975)

+ 0.6(0.08 − 0.082)(0.05 − 0.04975)
+ 0.25(0.10 − 0.082)(0.055 − 0.04975)
= 0.0000561

Correlation(Ri, Rj)
Covariance(R ABC , RX YZ )
=
Standard deviation(RABC) × Standard deviation(RXYZ)

Therefore:

0.0000561
Correlation = = 0.976
(0.01249 × 0.0046)

The correlation between the returns of the two companies is very strong (almost +1), and the

returns move linearly in the same direction.

Example: Calculating Correlation Coefficient #3

An analyst studied five years of historical data to examine how changes in Central Bank interest

132
© 2014-2024 AnalystPrep.
rates affect the country's inflation rate. The covariance between the interest rate and inflation

rate is -0.00075. The standard deviation of the interest rate is 5.5%, and the inflation rate is

12%. Now, let's calculate and interpret the correlation between these two variables.

Solution

CorrelationInterest rate, Inflation

CovarianceInterest Rate, Inflation
=
Standard deviation Interest rate × Standard deviationInflation
−0.00075
CorrelationInterest rate, Inflation = = −0.11364
(0.055 × 0.12)

A correlation of -0.11364 indicates a negative correlation between the interest rate and the

inflation rate.

Note that if we consider, say, assets A and B , then:

Cov(A, B)
Corr(A, B) = ρ(A, B) =
σA σB
⇒ Cov(A, B) = σA σBρ(A, B)

Consequently, in the formula for calculating portfolio variance, consisting of two assets, A and B,

we substitute for Cov(A, B) so that:

Portfolio Variance = W A2σ 2 (RA ) + W B2σ 2 (RB ) + 2(WA)(W B)σAσB ρ(A, B)

133
© 2014-2024 AnalystPrep.
Question

Assume that we have investments in two companies, ABC and XYZ. For ABC, there's a

15% chance of a 6% return, a 60% chance of an 8% return, and a 25% chance of a

10% return. The expected return for ABC is 8.2%, and the standard deviation is

1.249%. For XYZ, there are similar probabilities of 4%, 5%, and 5.5% returns. The

expected return for XYZ is 4.975%, and the standard deviation is 0.46%.

Assuming equal weights, the portfolio standard deviation is closest to:

A. 0.0000561.

B. 0.00007234.

C. 0.00851.

The correct answer is C.

Portfolio Variance = WA2 σ 2(R A) + WB2 σ 2(R B) + 2(W A)(WB )Cov(R A, R B)

First, we must calculate the covariance between the two stocks:

Cov(RABC ,R XY Z ) = 0.15(0.06 − 0.082)(0.04 − 0.04975)

+ 0.6(0.08 − 0.082)(0.05 − 0.04975)
+ 0.25(0.10 − 0.082)(0.055 − 0.04975)
= 0.0000561

Since we already have the weight and the standard deviation of each asset, we can

proceed and calculate the portfolio variance:

Portfolio variance = 0.52 × 0.012492 + 0.52 × 0.00462

+ 2 × 0.5 × 0.5 × 0.0000561
= 0.00007234

Therefore, the standard deviation is:

√0.00007234 = 0.00851

134
© 2014-2024 AnalystPrep.
LOS 5b: Calculate and interpret the covariance and correlation of
portfolio returns using a joint probability function for returns

Historical covariance or other techniques, such as market model regression with historical

return data, can help us forecast return covariance and correlation. We use the joint probability

function of the random variables for this estimation.

The probability that values of the two random variables X and Y will occur simultaneously is

given by the joint probability function of X and Y , denoted as P (X, Y ). For instance,

P (X = 3, Y = 4) represents the likelihood that X and Y will be equal to 3 and 4, respectively.

Covariance can be defined as a probability-weighted average of the cross-products of each

random variable's deviation from its own expected value. That is

Cov (Xi Yi ) = E[(Xi − X̄)(Yj − Y¯)]

= ∑ ∑ P (X = x i , Y = yj )(Xi − X̄ )(Y j − Y¯ )
i j

This formula calculates the covariance between random variables X and Y, such as portfolio

returns.

To find it, we take the sum of the products of the deviations of X and Y from their expected

values for all possible outcomes.

Each product is weighted by the probability of that specific outcome occurring.

Independence and Correlation

Two random variables, X and Y , are independent if P (X, Y ) = P (X) ⋅ P (Y ) . That is, X and Y are

independent. We find the product of independent probability to calculate joint probability.

The independence property is stronger than correlation because the correlation coefficient

addresses linear relationships.

If random variables X and Y are uncorrelated (also holds for independent random variables),

135
© 2014-2024 AnalystPrep.
then:

E(XY ) = E(X) ⋅ E(Y )

Example: Calculating the Covariance #1

Assume we wish to find the variance of each asset and the covariance between the returns of

ABC and XY Z, given that the amount invested in each company is $1,000.

This table is used to calculate the expected returns:

Strong Normal Weak

Economy Economy Economy
Probability 15% 60% 25%
ABC Returns 40% 20% 0%
XYZ Returns 20% 15% 4%

Solution

For us to find the covariance, we must calculate the expected return of each asset as well as

their variances. The assets’ weights are:

1000
W ABC = = 0.5
2000
1000
WXY Z = = 0.5
2000

Next, we should calculate the individual expected returns:

E(R ABC ) = 0.15 × 0.40 + 0.60 × 0.2 + 0.25 × 0.00 = 0.18

E(RX YZ ) = 0.15 × 0.2 + 0.60 × 0.15 + 0.25 × 0.04 = 0.13

Finally, we can compute the covariance between the returns of the two assets:

Cov(R ABC,XY Z ) = 0.15(0.40 − 0.18)(0.20 − 0.13)

+ 0.6(0.20 − 0.18)(0.15 − 0.13)
+ 0.25(0.00 − 0.18)(0.04 − 0.13)
= 0.0066

Example: Calculating the Covariance #2

136
© 2014-2024 AnalystPrep.
A portfolio manager is considering the following two possible economic growth of a country and

the joint variability of returns on two stocks in a portfolio:

Economic Growth < 4% > 4%

Probability 40% 60%
Return of Stock A 2.3% 8%
Return of Stock B 6.5% 3%

What is the covariance between the return of Stock A and Stock B?

Solution

Expected return of Stock A = (40% × 2.3%) + (60% × 8%) = 5.72%

Expected return of Stock B = (40% × 6.5%) + (60% × 3%) = 4.40%

Note: For the rest of the calculation, your curriculum sometimes ditches the percentage signs so

that 4.40% becomes simply 4.40.

The deviations of returns at the economic growth of

< 4% = (2.3 − 5.72) × (6.5 − 4.40) =?7.182

The deviations of returns at the economic growth of

> 4% = (8 − 5.72) × (3 − 4.40) = ?3.192

The covariance of returns between stock A and stock B is computed as follows:

Cov(RA ,B) = (−7.182 × 0.40) + (−3.192 × 0.60) = −4.788

Since covariance is negative, the two returns show some co-movement in opposite signs.

137
© 2014-2024 AnalystPrep.
Question

The following table represents the estimated returns for two motor vehicle

production brands – TY and Ford, in 3 industrial environments: strong (50%

probability), average (30% probability), and weak (20% probability).

TY Returns TY Returns Y Returns

+6% +3% −1%
Ford Sales + 10% Strong(0.5)
Ford Sales + 4% Average(0.3)
Ford Sales − 4% Weak(0.2)

Given the above joint probability function, the covariance between T Y and Ford

returns is closest to:

A. 0.054.

B. 0.1542.

C. 0.1442.

Solution

The correct answer is C.

First, we must start by calculating the expected return for each brand:

The expected return for TY:

= (0.5 × 6%) + (0.3 × 3%) + (0.2 × (−1%))

= 3% + 0.9% − 0.2% = 3.7%

The expected return for Ford:

= (0.5 × 10%) + (0.3 × 4%) + (0.2 × (−4%))

= 5% + 1.2% − 0.8% = 5.4%

Next, we can now compute the covariance:

138
© 2014-2024 AnalystPrep.
 Covariance = 0.5(6% − 3.7%)(10% − 5.4%)
+ 0.3(3% − 3.7%)(4%– 5.4%)
+ 0.2(−1% − 3.7%)(−4%– 5.4%)
= 5.29% + 0.294% + 8.836%
= 0.1442

The covariance is positive. This means that the returns for the two brands show some

co-movement in the same direction.

In real life, this scenario is highly likely because the companies belong to the same

industry. As a result, they share similar systematic risks.

139
© 2014-2024 AnalystPrep.
LOS 5c: define shortfall risk, calculate the safety-first ratio, and identify
an optimal portfolio using Roy’s safety-first criterion

Modern Portfolio Theory (MPT) evaluates investment options based on mean return and return

variance. This approach is applicable when investors are risk-averse, meaning they seek to

maximize their expected satisfaction or utility from their investments.

Mean-return analysis holds under two assumptions:

i. Returns follow a normal distribution.

ii. Investors have quadratic utility functions, a mathematical model representing the

balance between risk and return.

The mean-variance analysis can be reasonably accurate even if the two assumptions aren't

entirely met. Professionals prefer using observable data, such as returns. The assumption that

returns roughly follow a normal distribution has played a crucial role in applying MPT.

Mean-variance analysis only considers risk symmetrically. This implies that standard deviation

reflects variability above and below the mean. An alternative strategy is focusing on downside

risk. One such method is safety-first rules.

Before we dive into safety-first rules, let's discuss shortfall risk.

Shortfall Risk

Shortfall risk refers to the probability that a portfolio will not exceed the minimum (benchmark)

return an investor sets. In other words, it is the risk that a portfolio will fall short of the level of

return considered acceptable by an investor. As such, shortfall risks are downside risks. While a

shortfall risk focuses on the downside economic risk, the standard deviation measures the overall

volatility of a financial asset.

140
© 2014-2024 AnalystPrep.
Safety-First Ratio

Roy's safety-first criterion states that the optimal portfolio is the one that minimizes the

probability that a portfolio return, denoted by RP , may fall below the threshold level of return, RL

. The optimal portfolio minimizes P (R P < RL )

As such, if returns are distributed normally, the optimal portfolio is the one with the highest

safety-first ratio defined as:

E(R P ) − R L
SFRatio =
σP

141
© 2014-2024 AnalystPrep.
The numerator, E(RP − RL ), represents the distance from the mean return to the threshold level,

i.e., it measures the excess return over and above the threshold level of return per unit risk.

Intuitively, if the returns are normally distributed, the safety-first optimal portfolio maximizes the

SFRatio.

Given a portfolio SFRatio, the probability that its return will be less than RL is:

P (RP < R L ) = N(−SFRatio)

The safety-first optimal portfolio has the lowest P (RP < RL ).

Example: Safety-first Ratio

An investor sets a minimum threshold of 3%. There are three portfolios from which he is to

choose one. The expected return and the standard deviation for each portfolio are given below:

Portfolio A Portfolio B Portfolio C

Expected return 5% 10% 20%
Standard deviation 15% 20% 25%

What is the optimal portfolio for the investor?

Solution

Compute the safety-first ratio for each of the three portfolios and then compare them.

For portfolio A:

5− 3
SFRatio A = = 0.1333
15

Similarly, for portfolio B:

10 − 3
SFRatio B = = 0.35
20

Lastly:

20 − 3
142
© 2014-2024 AnalystPrep.
 20 − 3
SFRatioC = = 0.68
25

The optimal portfolio should maximize the safety-first ratio. Comparing the three ratios, it is easy

to notice that the safety-first ratio for portfolio C is the highest. Therefore, the investor should

choose portfolio C.

143
© 2014-2024 AnalystPrep.
Question

The returns on a fund are distributed normally. At the end of year t, the fund has a

value of $100,000. At the end of year t + 1, the fund manager wishes to withdraw

$10,000 for further funding but is reluctant to tap into the $100,000. There are two

investment options:

Portfolio A Portfolio B
Expected return 14% 13%
Standard deviation 17% 20%

Which portfolio is preferable for the manager?

A. Portfolio A.

B. Portfolio B.

C. The manager is indifferent to the two portfolios.

Solution

The correct answer is A.

First, you should calculate the threshold return from the information given. Since

there should be no tapping into the fund, the threshold return is:

10 , 000
= 10% or 0.1
100, 000

You should then calculate the safety-first ratio for each portfolio:

14 − 10
SFRatioA = = 0.24
17
13 − 10
SFRatioB = = 0.15
20

Portfolio A has the highest safety-first ratio. This is the reason it is the most desirable.

You can also go a step further and calculate P (R P < R L ). To do this, you would have to

144
© 2014-2024 AnalystPrep.
negate each safety-first ratio and then find the CDF of the standard normal

distribution for the resulting value. That is,

P (R P < R L ) = N (−SFRatio)
N (−0.24) = 1 − N (0.24)
= 1 − 0.5948 = 0.4052
N (−0.15) = 1 − N (0.15)
= 1 − 0.5596 = 0.4404

(−where SFRatio is the z-value)

For portfolio A, there is approximately a 40% probability of obtaining a return below

the threshold return. For portfolio B , this probability rises to 44%. Therefore, we

choose the option for which the chance of not exceeding the benchmark return is

lowest – portfolio A.

145
© 2014-2024 AnalystPrep.
 Learning Module 6: Simulation Methods

LOS 6a: explain the relationship between normal and lognormal

distributions and why the lognormal distribution is used to model asset
prices when using continuously compounded asset returns

A random variable Y is lognormally distributed if its natural logarithm, In Y , is normally

distributed. The opposite is true. If ln Y is normally distributed, then Y is lognormally

distributed.

The lognormal distribution is positively skewed, meaning it's skewed to the right and has a long

right tail. In this distribution, values are bounded by 0. Typically, the mean is greater than the

mode.

Consider the following graph of two probability density functions (pdfs) of two lognormal

distributions.

146
© 2014-2024 AnalystPrep.
Like the normal distribution, two parameters – the mean and variance of the associated normal

distribution – fully describe the lognormal distribution.

Expressions for Mean and Variance of Lognormal Distribution

Assume that X is normally distributed with the mean μ and variance σ 2. Also, define the variable

Y = eX .

Then ln Y = ln (eX ) = X is lognormally distributed with the following mean and variance

expressions:

147
© 2014-2024 AnalystPrep.
 1
(μ+ σ 2 )
Mean = μL = e 2

2 2
Variance = σL2 = e2u +σ (eσ − 1)

Why the Lognormal Distribution is Used to Model Stock Prices

The lognormal distribution works well for modeling asset prices that cannot be negative because

it has a lower bound at zero.

When the continuously compounded returns on a stock follow a normal distribution, the stock

prices follow a lognormal distribution. Note that even if returns do not follow a normal

distribution, the lognormal distribution is still the most appropriate for stock prices.

Continuously Compounded Rate of Return

Remember that given the investment horizon from time t = 0 to time t = T , the continuously

compounded return of a stock is given by:

PT
r0 ,T = ln ( )
P0

If we apply the exponential function on both sides of the equation, we have the following:

PT = P 0 er0,T

PT
Note that can be written as:
P0

PT PT P T−1 P
=( )( ) …( 1 )
P0 P T −1 P T−2 P0

If we take natural logarithm on both sides of the above equation:

PT PT P T−1 P
ln ( ) = ln (( )( ) … ( 1 ))
P0 PT −1 P T−2 P0
⇒ r0,T = rT −1,T + rT−2 ,T−1 + … + r0,1

148
© 2014-2024 AnalystPrep.
Therefore, the continuously compounded return to time T equals the sum of one-period

continuously compounded returns.

Remember that a linear combination of normal random variables is also normal. Therefore, if the

shorter period returns rT −1,T , rT −2 ,T −1 , … , r0 ,1 are normally distributed or approximately normal,

then r0,T will also be approximately normal.

Furthermore, if we assume that the one-period continuously compounded returns

rT−1 ,T , rT −2,T −1 , … , r0,1 are independently and identically distributed (i.i.d) random variables with

a mean of μ and variance of σ 2, then:

The expected value of the continuously compounded return over a holding period of T

periods is given by:

E (r0 ,T ) = E (rT −1,T ) + E (rT −2 ,T −1 ) + … + E (r0,1 ) = μT

The variance of the continuously compounded return over a holding period is given by:

σ 2 (r0 ,T ) = σ 2T

The standard deviation of the continuously compounded returns, also known as volatility, is given

by:

σ(r0, T ) = σ√ T

In other words, if rT−1 ,T , rT −2, T−1 , … , r0, 1 are normally distributed with the mean of μ and variance

of σ 2 then r0 ,T is normally distributed with the mean of μT and variance of σ 2T .

Let us go back to the formula:

PT = P 0 er0,T

If X is normally distributed with the mean μ and variance σ2 and that Y = eX then,

ln Y = ln (eX) = X is lognormally distributed. Assuming we apply this intuition in the above

formula, it would be easy to see that we can model P T as a lognormally distributed random

149
© 2014-2024 AnalystPrep.
variable since r0 ,T is approximately normally distributed.

Volatility and Continuously Compounded Returns

Volatility measures the standard deviation of the continuously compounded returns on the

underlying asset. Conventionally, it is usually annualized.

We calculate volatility using the historical series of continuously compounded returns. Another

method is converting daily holding returns into continuously compounded daily returns and then

calculating annualized volatility.

We base annualizing volatility on 250 trading days in a year, which is an estimate of the business

days the financial markets operate. The formula we use for annualizing volatility is:

σ(r0, T ) = σ√T

For example, if the daily volatility is 0.05, then the annual volatility is:

σ(r0 ,T ) = 0.05 × √250 = 0.79

Example: Lognormal Distribution and Continuous Compounding

Jess Kasuku is analyzing the stock of ABC Company, which is listed on the London Stock

Exchange under the ABC ticker symbol. Kasuku wants to understand how the stock's price

changed during a particular week when significant developments in the global economy

impacted the UK stock market. To do this, she calculates the stock's volatility for that week using

the closing prices shown in Table 1.

Table 1: ABC Company Daily Closing Prices

Day Closing Price (GBP)
Monday 75
Tuesday 78
Wednesday 72
Thursday 70
Friday 68

Using the information in Table 1, calculate the annualized volatility of ABC Company's stock for

150
© 2014-2024 AnalystPrep.
that week, assuming 250 trading days in a year.

Solution

Step 1: Calculate the continuously compounded daily returns for each day using the formula
Ending Price
ln( Beginning Price
):

78
r1 = ln ( ) = 0.03922
75
72
r2 = ln ( ) = −0.08004
78
70
r3 = ln ( ) = −0.02817
72
68
r4 = ln ( ) = −0.02899
70

Step 2: Calculate the mean of the continuously compounded daily returns:

r1 + r2 + r3 + r4
μ=
4
0.03922 + (−0.08004) + (−0.02817) + (−0.02899)
=
4
= −0.024495

Step 3: Calculate the variance of the continuously compounded daily returns:

(r1 − μ)2 + (r2 − μ)2 + (r3 − μ)2 + (r4 − μ)2

σ2 =
4
[(0.03922 − (−0.24495))2 + (−0.08004 − (−0.024495))2

+ (−0.02817 − (−0.024495))2 + (−0.02899 − (−0.024495))2 ]

=
4
0.007179
= = 0.001795
4

Step 4: Calculate the standard deviation of the continuously compounded daily returns:

σ = √σ 2
= √ 0.001795 = 0.042363

Step 5: Annualize the volatility by multiplying the daily volatility by the square root of the

151
© 2014-2024 AnalystPrep.
number of trading days in a year.

We know that:

σ(r0 ,T ) = σ√T
∴ σannualized = σdaily × √250
= 0.042363 × √ 250
= 0.6698 ≈ 67%

So, the annualized volatility of ABC Company's stock for that week was 67.23 percent.

152
© 2014-2024 AnalystPrep.
Question

Which of the following is true about lognormal distributions compared to normal

distributions?

A. They are skewed to the right.

B. They can take on negative values.

C. They are less suitable for describing asset prices than asset returns.

Solution

The correct answer is A.

Lognormal distributions are continuous probability distributions that only take

positive values and are often skewed to the right.

B is incorrect because lognormal distributions only take on positive values.

C is incorrect because there is no evidence to suggest that lognormal distributions

are less suitable for describing asset prices than asset returns.

153
© 2014-2024 AnalystPrep.
LOS 6b: describe Monte Carlo simulation and explain how it can be used
in investment applications

Monte Carlo simulations are about producing many random variables based on specific

probability distributions. This helps in estimating the probability of various results.

We will give an example to illustrate Monte Carlo Simulation implementation.

Steps Involved in Project Appraisal

Imagine an investor who wants to predict the results of a 70% stock and 30% bond portfolio over

20 years. This is how we set up a Monte Carlo simulation:

Specifying the Simulation

Step 1: Specify the quantity of interest in terms of underlying variables

The quantity of interest here could be the final portfolio value after 20 years, denoted as ViT . In

this case, this is the final portfolio value at time T resulting from ith simulation trial.

The underlying variable is the return on the portfolio. The starting portfolio value is $100,000,

with 70% invested in stocks and 30% in bonds.

Step 2: Specify a time horizon

Assume we're interested in yearly returns, so the time horizon is 20 years. Divide the calendar

time into sub-periods. In this case, we will assume yearly returns so that the number of sub-

periods is K = 20, and the time increment Δ t is, therefore, one year.

Step 3: Specify the method for generating the data used in the simulation

Here, we need to make distributional assumptions. We might assume that the annual portfolio

return follows a normal distribution. Let's say we estimate an average return μ of 7% for stocks,

3% for bonds, a standard deviation σ of 15% for stocks, and 5% for bonds. We can model changes

in the portfolio value using the formula below:

154
© 2014-2024 AnalystPrep.
 ΔPortfolio value = 0.7 ∗ (μ stock × Prior portfolio value × Δt
+ σstock × Prior portfolio value × Zk )
+ 0.3 ∗ (μ bond × Prior portfolio value × Δt
+ σbond × Prior portfolio value × Zk )

Here, Zk is a standard normal random variable representing the uncertainty in the portfolio

return (risk factor). We can use a computer program to draw 20 random values of Zk .

Running the Simulation Over a Given Number of Trials

Step 4: Use the simulated values to produce portfolio values

This step involves converting the standard normal random numbers (Zk) generated in step 3 into

yearly changes in portfolio value (ΔPortfolio value) using our model from step 3. This gives us 20

observations of possible changes in portfolio value over the 20-year period. From these

observations, we create a sequence of 20 portfolio values, starting with the initial value of

$100,000.

Step 5: Calculate the final portfolio value

The average portfolio value at the end of 20 years (V iT ) is calculated by summing up the portfolio

values at the end of each year and dividing by 20. We then calculate the present value (Vi0 ) of

this average value by discounting it to the present using an appropriate interest rate. The

subscript i in V iT and Vi0 indicates that these values are from the ith simulation trial. This

completes one simulation trial.

Step 6: Repeat steps 4 and 5 over the required number of trials

Finally, we repeat steps 4 and 5 multiple times, say, 1,000 times. We then calculate summary

statistics, such as the mean, median, and percentiles of the distribution of Vi 0 values. These

summary statistics provide a range of potential outcomes for the portfolio value after 20 years,

helping the investor understand the risks and rewards of the investment strategy.

Major Applications of Monte Carlo Simulations

155
© 2014-2024 AnalystPrep.
 It can also be used to value complex securities such as American or European options.

Limitations of Monte Carlo Simulations

It only provides us with statistical estimates of results, not exact figures.

It is fairly complex and can only be carried out using specially designed software that

may be expensive.

The complexity of the process may cause errors, leading to wrong results that can be

potentially misleading.

156
© 2014-2024 AnalystPrep.
Question

Which of the following is a correct statement about the use of Monte Carlo

simulations in finance and investment?

A. They provide exact valuations of call options.

B. They estimate a portfolio's potential returns by simulating its performance.

C. They assess how changes in assumptions, such as interest rates or market

volatility, affect a financial model.

Solution

The correct answer is C.

Monte Carlo simulations can assess how changes in assumptions, such as interest

rates or market volatility, affect a financial model. This allows analysts to understand

the impact of these changes on the model's results.

A is incorrect because Monte Carlo simulations do not provide exact valuations of

call options. Instead, they can estimate the value of these options by simulating their

potential outcomes.

B is incorrect because while Monte Carlo simulations can estimate a portfolio's

potential returns, they do not simply simulate its performance. Instead, they use

probability distributions to model the uncertainty in the portfolio's returns.

157
© 2014-2024 AnalystPrep.
LOS 6c: describe the use of bootstrap resampling in conducting a
simulation based on observed data in investment applications

Resampling

Resampling means repeatedly drawing samples from the original observed sample to make

statistical inferences about population parameters. There are two common methods: Bootstrap

and jackknife. Here, we'll focus on the Bootstrap method.

Bootstrap Resampling

Bootstrap resampling relies on computer simulations for statistical inferences, bypassing the

need for conventional analytical formulas like z-statistics. The bootstrap technique is

underpinned by a strategy that mirrors the random sampling process from a population to create

a sampling distribution.

158
© 2014-2024 AnalystPrep.
Note that in bootstrapping, we do not have information about the population. Our only insight

comes from a sample of size n drawn from this “unknown population.”

The core concept is that a random sample can effectively stand in for the entire population. So,

we can mimic drawing samples from the population by repeatedly resampling from the initial

sample. Essentially, the bootstrap method treats the initially obtained sample as a stand-in for

the entire population.

Bootstrapping vs Monte Carlo Simulation

Both bootstrap and Monte Carlo simulation techniques heavily rely on repetitive sampling.

Bootstrap considers the resampled dataset a proxy for the true population and infers population

parameters such as mean, variance, skewness, and kurtosis from the statistical distribution of

these samples.

159
© 2014-2024 AnalystPrep.
Conversely, Monte Carlo simulation is centered on the generation of random data with a pre-

determined statistical distribution of parameter values.

Simulation Using Bootstrapping

Simulation using bootstrapping is similar to Monte Carlo Simulation except for the source of

random variables. In bootstrapping, the random variables are taken from a bootstrap sample

instead of a probability distribution.

Consider the previous example:

Let's say an investor wants to understand the potential outcomes of investing in a portfolio with

a 70-30 split between stocks and bonds over a 20-year period. Here's how a Monte Carlo

simulation could be set up.

The simulation steps using the bootstrap sampling distribution are as follows:

Specifying the Simulation

Step 1: Specify the quantity of interest in terms of underlying variables

The quantity of interest here could be the final portfolio value after 20 years, denoted as ViT . The

underlying variable is the return on the portfolio. The starting portfolio value is $100,000, with

70% invested in stocks and 30% in bonds.

Step 2: Specify a time horizon

Assume we're interested in yearly returns, so the time horizon is 20 years. Divide the calendar

time into sub-periods. In this case, we will assume yearly returns so that the number of

subperiods is K = 20, and time increment Δt is, therefore, one year.

Step 3: Generate bootstrap samples from the empirical distribution of portfolio returns

Here, we use the historical return data as our empirical distribution. Instead of assuming that

the annual portfolio return follows a specific theoretical distribution, we will use the bootstrap

procedure to draw the K = 20 yearly returns from the observed empirical distribution.

160
© 2014-2024 AnalystPrep.
Running the Simulation Over a Given Number of Trials

Step 4: Use the bootstrap samples to produce portfolio values used to value the

contingent claim

This step involves using the bootstrap samples drawn in Step 3 to compute the yearly changes in

portfolio value. From there, we create a sequence of 20 portfolio values, starting with the initial

value of $100,000.

Step 5: Calculate the final portfolio value

The average portfolio value at the end of 20 years (V iT ) is calculated by summing up the portfolio

values at the end of each year and dividing by 20. We then calculate the present value (Vi0 ) of

this average value by discounting it to the present using an appropriate interest rate. The

subscript i in V iT and Vi0 indicates that these values are from the ith bootstrap sample. This

completes one bootstrap sample.

Step 6: Repeat steps 4 and 5 over the required number of trials

Finally, we repeat steps 4 and 5 multiple times, say, 1,000 times. We then calculate summary

statistics, such as the mean, median, and percentiles of the distribution of Vi 0 values. These

summary statistics provide a range of potential outcomes for the portfolio value after 20 years,

helping the investor understand the risks and rewards of the investment strategy based on the

observed empirical distribution of returns.

161
© 2014-2024 AnalystPrep.
Question

Which of the following statements is most likely accurate in relation to bootstrap

analysis?

A. Bootstrap analysis aims to deduce statistics about population parameters

from a singular sample.

B. Bootstrap analysis involves the repeated extraction of samples of equal size,

with replacement, from the initial population.

C. During bootstrap analysis, it is necessary for analysts to determine

probability distributions for primary risk factors that govern the underlying

random variables.

Solution

The correct answer is A.

The bootstrap analysis employs random sampling to generate an observed variable

from a set of unknown population parameters. Although the actual distribution of the

population is unknown to the analyst, the parameters of the population can be

inferred through the sample produced via random sampling.

B is incorrect. In bootstrap analysis, the analyst repeatedly samples from the initial

sample, not the entire population. Each resample has the same size as the original

sample, and for each new draw, selected items go back into the sample.

C is incorrect. During bootstrap analysis, analysts simply utilize the empirical

distribution of the observed underlying variables. In contrast, the analyst must

establish probability distributions for the key risk factors that govern the underlying

variables in a Monte Carlo simulation.

162
© 2014-2024 AnalystPrep.
 Learning Module 7: Estimation and Inference

LOS 7a: compare and contrast simple random, stratified random, cluster,
convenience, and judgmental sampling and their implications for
sampling error in an investment problem

Sampling is the systematic process of selecting a subset or sample from a larger population.

Sampling is essential because it is costly and time-consuming to analyze the whole population.

Sampling methods can be broadly categorized into probability sampling and non-probability

sampling.

In probability sampling, every population member has an equal chance of being chosen for the

sample, ensuring a representative sample. In contrast, non-probability sampling depends on

factors such as the sampler's judgment or data accessibility, increasing the risk of an

unrepresentative sample.

Probability Sampling Methods

Simple Random Sampling

Simple random sampling means selecting a sample from a population where each element has an

equal chance of being chosen. This method aims to create an unbiased sample that accurately

represents the population.

Simple random sampling is appropriate when applied to a homogeneous population.

Example: Simple Random Sampling

Imagine we wish to come up with a sample of 50 CFA level I candidates out of 100,000.

One approach may involve numbering each of the 100,000 candidates, placing them in a basket,

and shaking the basket to jumble up the numbers. Next, we would randomly draw 50 numbers

from the basket, one after the other, without replacement.

163
© 2014-2024 AnalystPrep.
A more scientific approach may also involve the use of random numbers. All the 100,000

candidates are numbered in a sequence (from 1 to 100,000). We may then use a computer to

randomly generate 50 numbers between 1 and 100,000, where a given number represents a

particular candidate whose name or admission number can be identified.

The underlying feature of random sampling is that all elements in the population must have

equal chances of being chosen.

Stratified Random Sampling

In stratified random sampling, analysts subdivide the population into separate groups known as

strata (singular stratum). Each stratum comprises elements with a common characteristic

(attribute) that distinguishes them from all the others. The method is most appropriate for large

heterogeneous populations.

A simple random sample is then drawn from within each stratum and combined to form the

overall, final sample that takes heterogeneity into account. The number of members chosen from

any one stratum depends on its size relative to the population as a whole.

Example: Stratified Random Sampling

An advertising firm wants to determine the extent to which it needs to invigorate television

advertisements in a district. The company decides to conduct a survey to estimate the mean

number of hours households spend watching TV per week. The district has three distinct towns –

A, B, which are urbanized, and C , located in a rural area. Town A is adjacent to a major factory

where most residents work, with most having kids of school-going age. Town B mainly harbors

retirees, while most people in town C practice agriculture.

There are 160 households in town A, 60 in town B, and 80 in C . Given the differences in the

composition of each region, the firm decides to draw a sample of 50 households, considering the

total number of families in each.

What is the number of homes that have been sampled in each town?

Solution

164
© 2014-2024 AnalystPrep.
We have three strata: towns A, B, and C. We use the following formula to determine the number

of households from each region to be included in the sample:

Number of households in sample

Number of households in the region
=( ) × Required sample size
Total number of households

Therefore, the number of households to be sampled in town A,

160
= × 50 = 27 (approximately)
300

Similarly, the number of households to be sampled in town B,

60
= × 50 = 10
300

80
Finally, the firm would need ( 300 × 50) = 13 households in town C.

Advantages of Stratified Sampling over Simple Random Sampling

Stratification is associated with a smaller estimation error compared to simple random

sampling, especially when each stratum is homogeneous.

Stratification enables analysts to estimate the population parameter, say, the mean for

all the subgroups of the entire population.

Cluster Sampling

Cluster sampling involves categorizing all population elements into distinct and all-encompassing

groups called clusters. Then, you can either choose a random sample of entire clusters or select

a random subset from each cluster. So, there are two cluster sampling approaches:

One-stage (or single-stage) cluster sampling: All the members in each sampled cluster

are sampled.

Two-stage cluster sampling: A simple random sub-sample of members is selected from

165
© 2014-2024 AnalystPrep.
 each cluster.

Key Point Difference Between Stratified Sampling and Cluster Sampling

In cluster sampling, a cluster serves as a single sampling unit, and only specific

clusters are sampled.

In stratified sampling, you select members from within each stratum and then draw a

random sample from each stratum.

Non-Probability Sampling Techniques

Non-probability samples are selected based on judgment or the convenience of accessing data.

As such, non-probability sampling depends on the researchers' sample selection skills. There are

two types of non-probability sampling methods:

i. Convenience sampling: Researchers choose population elements based on ease of

access. This method may not provide a fully representative sample, limiting sampling

accuracy.

ii. Judgmental sampling: Researchers select elements subjectively, often based on their

own knowledge and expertise. However, this approach can introduce bias and result in a

166
© 2014-2024 AnalystPrep.
 non-representative sample.

Judgmental sampling is preferred when a restricted number of people in a population possess

qualities that the researcher expects from the target population.

Comparison Between Probability Sampling and Nonprobability Sampling

Method Strengths Weaknesses

Probability Sampling
Simple random sampling Easy to use Lower precision;
no assurance of
representativeness
Stratified sampling Higher precision Difficult to choose
relative to simple relevant stratification;
random sampling expensive
Cluster sampling Cost-effective and Lower precision
efficient
Non-probability
Sampling
Convenience sampling Cost-effective and Selection bias, sample
saves time; easy may not accurately
to use represent population
Judgmental sampling Cost-effective, Subjective method.
convenient, less time Selection bias, sample
consuming may not accurately
represent the population.

Sampling Error and Its Implications on Investment

Sampling error refers to the difference between the observed value (results obtained from

analyzing a sample of investment data) and the true values that would have been obtained from

analyzing the entire population of investments.

For instance, when we take a sample to estimate a population's mean, there's typically a

difference between the sample mean and the true population mean. This difference, known as

sampling error, emerges due to natural variation in sampling and because we work with data

from only a part of the full population.

Therefore, any conclusions or predictions drawn based on the sample data may deviate from the

actual performance or characteristics of the entire investment population.

167
© 2014-2024 AnalystPrep.
Question 1

An analyst is analyzing the spending habits of people belonging to different annual

income categories. In his analysis, he creates the following different groups

according to the annual family income: Less than $30,000, $31,000 – $40,000,

$41,000 to $50,000, and $51,000 to $60,000. He then selects a sample from each

distinct group to form a whole sample. The sampling method used by the analyst is

most likely:

A. Cluster sampling.

B. Stratified sampling.

C. Simple random sampling.

Solution

The correct answer is B.

Dividing the population into different strata/groups and selecting a sample from each

group is called the stratified sampling technique.

A is incorrect. In cluster sampling, each cluster is considered a sampling unit, and

only selected clusters are sampled.

C is incorrect. Simple random sampling involves the selection of a sample from an

entire population such that each member or element of the population has an equal

probability of being picked.

Question 2

A Ph.D. student is conducting research related to her thesis, and for this purpose, she

uses some students from her university to constitute a sample. The sampling method

used by the analyst is most likely:

168
© 2014-2024 AnalystPrep.
 A. Simple random sampling.

B. Convenience sampling.

C. Judgmental sampling.

Solution

The correct answer is B.

The researcher has selected the students from her university because she can

conveniently access them.

A is incorrect. Simple random sampling involves the selection of a sample from an

entire population such that each member or element of the population has an equal

probability of being picked.

C is incorrect. Judgmental sampling involves handpicking elements from a sample

based on the researcher's knowledge and expertise.

Question 3

An analyst wants to estimate the downtime of ABC Bank's ATMs in a city for the last

six months. For this purpose, he selects 20 locations or areas within the city and then

selects 50% of the ATMs in each area. The sampling method used by the analyst is

most likely:

A. Cluster sampling.

B. Stratified random sampling.

C. Simple random sampling.

Solution

The correct answer is A.

169
© 2014-2024 AnalystPrep.
In cluster sampling, all population elements are categorized into mutually exclusive

and exhaustive groups called clusters. A simple random sample of the cluster is

selected, and then the elements in each of these clusters are sampled.

B is incorrect. In stratified random sampling, analysts subdivide the population into

separate groups known as strata (singular–stratum), and each stratum is composed of

elements that have a common characteristic (attribute) that distinguishes them from

all the others.

C is incorrect. Simple random sampling involves the selection of a sample from an

entire population such that each member or element of the population has an equal

probability of being picked.

170
© 2014-2024 AnalystPrep.
LOS 7b: explain the central limit theorem and its importance for the
distribution and standard error of the sample mean

The central limit theorem asserts that “given a population described by any probability

distribution having mean μ and finite variance σ 2, the sampling distribution of the sample mean

X̄ computed from random samples of size n from this population will be approximately normal
2
σ
with mean μ (the population mean) and variance (the population variance divided by n ) when
n

the sample size n is large”.

What is a Large Enough n?

The answer to this question might not be straightforward. Nevertheless, the widely accepted

value is n ≥ 30. The truth is that the value of n depends on the shape of the population involved,

i.e., the distribution of Xi and its skewness.

In a non-normal but fairly symmetric distribution, n = 10 can be considered large enough. With a

very skewed distribution, the value of n can be 50 or even more.

Standard Error of the Sample Mean

Remember that from the central limit theorem, the variance of the sample mean distribution is

given by:

σ2
σX̄2 =
n

Where n is the sample size.

The standard error is the standard deviation of the statistic (sample mean).

σ
σX̄ =
√n

Formally defined, for a sample mean X̄ computed from a sample generated by a population with

171
© 2014-2024 AnalystPrep.
standard deviation σ, the standard error of the sample mean is given by:

σ
σX̄ =
√n

Where σ = Known population standard deviation.

When the population standard deviation, σ, is unknown, the following formula is used to estimate

the standard error of the sample mean, also denoted as sX̄ :

s
sX̄ =
√n

Where: s = Sample standard deviation.

The formula above is applicable where we do not know the population standard deviation. Note

that the sample standard deviation is the square root of the sample variance, s2, given by:

2
∑ ni 1 (Xi − X̄ )
2 =
s =
n−1
2
∑ni=1 (Xi − X̄)
⇒s=⎷
n− 1

The standard error of the sample mean estimates the variation that would occur if you took

multiple samples from the same population. While the standard deviation measures variation

within one sample, the standard error estimates variation across many samples. So, standard

deviation and standard error are distinct concepts.

The standard error of the sample mean gives analysts an idea of how precisely the sample mean

estimates the population mean. A lower standard error value indicates a more precise estimation

of the population mean. On the other hand, a larger standard error value indicates a less precise

estimate of the population mean.

It is also important to note that the standard error becomes smaller as the sample size increases.

This can be seen from its formula. This happens because increasing the sample size ultimately

brings the sample mean closer to the true value of the population mean.

172
© 2014-2024 AnalystPrep.
Example 1

In a certain property investment company with an international presence, workers have a mean

hourly wage of $12 with a population standard deviation of $3. Given a sample size of 30, the

standard error of the sample mean is closest to:

σ
σX̄ =
√n
3
= = $0.55
√30

If we were to draw several samples of size 30 from the employee population and construct a

sampling distribution of the sample means, we would end up with a mean of $12 and a standard

error of $0.55.

Example 2

A sample of 30 latest returns on XYZ stock reveals a mean return of $4 with a sample standard

deviation of $0.13. The standard error of the sample mean is closest to:

s
sX̄ =
√n
0.13
= $0.02
√30

If we were to draw more samples from the population of yearly returns on XYZ stock and

construct a sample mean distribution, we would end up with a mean of $4 and a standard error

of $0.02.

173
© 2014-2024 AnalystPrep.
Question

Emma Johnson wants to know how finance analysts performed last year. Johnson

assumes that the population cross-sectional standard deviation of finance analyst

returns is 8 percent and that the returns are independent across analysts.

The random sample size that Johnson needs if she wants the standard deviation of the

sample means to be 2% is closest to:

A. 4.

B. 16.

C. 72.

Solution

The correct answer is B.

Remember that,

σ
σX̄ =
√n
0.08
⇒ 0.02 =
√n
∴ n = 16

174
© 2014-2024 AnalystPrep.
LOS 7c: describe the use of resampling (bootstrap, jackknife) to estimate
the sampling distribution of a statistic

Resampling refers to the act of repeatedly drawing samples from the original observed data

sample for the statistical inference of population parameters. The two commonly used methods

of resampling are bootstrap and jackknife.

Bootstrap

Using a computer, the bootstrap resampling method simulates drawing multiple random samples

from the original sample. Each resample is the same size as the original sample. These

resamples are used to create a sampling distribution.

175
© 2014-2024 AnalystPrep.
In the bootstrap method, the number of repeated samples drawn is at the researcher's

discretion. Note that bootstrap resampling is done with replacement.

Furthermore, we can calculate the standard error of the sample mean. This is done by

resampling and calculating the mean of each sample. The following formula is used to estimate

the standard error.

B 2
1
sX̄ = (^ − )
⎷ B − 1 b=1 θ b θ̄
∑

Where:

sX̄ = Estimate of the standard error of the sample mean.

B = Number of resamples drawn from the original sample.

θ^b = Mean of a resample.

θ̄= Mean across all the resample means.

The bootstrap resampling method can also be used to estimate the confidence intervals for

statistics on other population parameters, such as the median.

Advantages of Bootstrap Resampling

1. No Reliance on Analytical Formulas: Bootstrap differs from traditional statistics

because it doesn't rely on an analytical formula for estimating distributions. This makes

it versatile for complex estimators and especially useful when analytical formulas are

unavailable.

2. Applicability to Complicated Estimators: Bootstrap is a simple yet powerful method

that effectively handles complicated estimators. It can handle a wide range of statistical

models, making it suitable for various applications in finance where complex estimations

are common.

3. Increased Accuracy: Bootstrap can enhance accuracy by creating multiple resampled

datasets and estimating population parameters on each. This helps understand estimator

176
© 2014-2024 AnalystPrep.
 variability and robustness, ultimately improving result accuracy.

Jackknife

Jackknife is a resampling method in which samples are drawn by omitting one observation at a

time from the original data sample. This process involves drawing samples without replacement.

For a sample size of n, we need n repeated samples. This method can be used to reduce the bias

of an estimator or to estimate the standard error and the confidence interval of an estimator.

177
© 2014-2024 AnalystPrep.
Question

Assume that you are studying the median height of 100 students in a university. You

draw a sample of 1000 students and obtain 1000 median heights. The mean across all

resample means is 5.8. The sum of squares of the differences between each sample
2
mean, and the mean across all resample means ∑B
b=1
(θ^b − θ̄) is 2.3.

The Estimate of the standard error of the sample mean is closest to:

A. 0.05.

B. 0.08.

C. 0.10.

Solution

The correct answer is A.

B 2
1
sX̄ = ⎷ ∑ (θ^b − θ̄)
B − 1 b=1
1
=√ × 2.3 = 0.04798 ≈ 0.05
1000 − 1

178
© 2014-2024 AnalystPrep.
 Learning Module 8: Hypothesis Testing

LOS 8a: Explain hypothesis testing and its components, including

statistical significance, Type I and Type II errors, and the power of a test

A hypothesis is an assumed statement about a population's characteristics, often considered an

opinion or claim about an issue. To determine if a hypothesis is accurate, statistical tests are

used. Hypothesis testing uses sample data to evaluate if a sample statistic reflects a population

with the hypothesized value of the population parameter.

Below is an example of a hypothesis:

“The mean return of small-cap stock is higher than that of large-cap stock.”

Hypothesis testing involves collecting and examining a representative sample to verify the

accuracy of a hypothesis. Hypothesis tests help analysts to answer questions such as:

Is bond type A more profitable than type B?

Does staff training lead to improved efficiency at the workplace?

Are motor vehicle insurance claims consistent with a lognormal distribution?

Procedure Followed During Hypothesis Testing

Whenever a statistical test is being performed, the following procedure is generally considered

ideal:

1. Statement of both the null and the alternative hypotheses.

2. Selection of the appropriate test statistic, i.e., what's being tested, e.g., the population

mean, the difference between sample means, or variance.

3. Specification of the level of significance.

4. A clear statement of the decision rule to guide the choice of whether to reject or approve

the null hypothesis.

5. Calculation of the sample statistic.

179
© 2014-2024 AnalystPrep.
 6. Arrival at a decision based on the sample results.

Step 1: Stating the Hypotheses

The Null vs. Alternative Hypothesis

The null hypothesis, denoted as H0 , signifies the existing knowledge regarding the population

parameter under examination, essentially representing the "status quo." For example, when the

U.S. Food and Drug Administration inspects a cooking oil manufacturing plant to confirm that the

cholesterol content in 1 kg oil packages doesn't exceed 0.15%, they might create a hypothesis

like:

H 0: Each 1 kg package has 0.15% cholesterol.

A test would then be carried out to confirm or reject the null hypothesis.

Typical statements of H 0 include:

H 0 : μ = μ0
H 0 : μ ≤ μ0
H 0 : μ ≥ μ0

Where:

μ = True population mean.

μ0 = Hypothesized population mean.

The alternative hypothesis, denoted as H a , contradicts the null hypothesis. Therefore, rejecting

the H0 makes H a valid. We accept the alternative hypothesis when the “status quo” is discredited

and found to be false.

Using our FDA example above, the alternative hypothesis would be:

H a: Each 1 kg package does not have 0.15% cholesterol.

One-tailed vs. Two-tailed Hypothesis Testing

180
© 2014-2024 AnalystPrep.
One-tailed Test

A one-tailed test (one-sided test) is a statistical test that considers a change in only one direction.

In such a test, the alternative hypothesis either has a < (less than sign) or > (greater than sign),

i.e., we consider either an increase or reduction, but not both.

A one-tailed test directs all the significance levels (α) to test statistical significance in one

direction. In other words, we aim to test the possibility of a change in one direction and

completely disregard the possibility of a change in the other direction.

If we have a 5% significance level, we shall allot 0.05 of the total area in one tail of the

distribution of our test statistic.

Examples: Hypothesis Testing

Let us assume we are using the standardized normal distribution to test the hypothesis that the

population mean equals a given value X . Further, let us assume we are using data from a sample

drawn from the population of interest. Our null hypothesis can be expressed as:

H0 : μ = X

If our test is one-tailed, the alternative hypothesis will test if the mean is either significantly

greater than X or significantly less than X , but NOT both.

Case 1: At the 95% Confidence Level

Ha : μ < X

The mean is significantly less than X if the test statistic is in the bottom 5% of the probability

distribution. This bottom area is known as the critical region (rejection region). We will reject the

null hypothesis if the test statistic is less than -1.645.

181
© 2014-2024 AnalystPrep.
Case 2: Still at the 95% Confidence Level

Ha 1 : μ > X

We would reject the null hypothesis only if the test statistic is greater than the upper 5% point of

the distribution. In other words, we would reject H 0 if the test statistic is greater than 1.645.

182
© 2014-2024 AnalystPrep.
A Two-tailed Test

A two-tailed test considers the possibility of a change in either direction. It looks for a statistical

relationship in both a distribution's positive and negative directions. Therefore, it allows half the

value of α to test statistical significance in one direction and the other half to test the same in the

opposite direction. A two-tailed test may have the following set of hypotheses:

H0 : μ = X
H1 : μ ≠ X

Refer to our earlier example. If we were to carry out a two-tailed test, we would reject H0 if the

test statistic turned out to be less than the lower 2.5% point or greater than the upper 2.5%

point of the normal distribution.

183
© 2014-2024 AnalystPrep.
Step 2: Identify the Appropriate Test Statistic and Distribution

Test Statistic

A test statistic is a standardized value computed from sample information when testing

hypotheses. It compares the given data with what an analyst would expect under a null

hypothesis. As such, the null hypothesis is a major determinant of the decision to accept or reject

H 0, the null hypothesis.

We use test statistics to gauge the degree of agreement between sample data and the null

hypothesis. Analysts use the following formula when calculating the test statistic for most tests:

Sample statistic − Hypothesized value

Test statistic =
Standard error

The test statistic is a random variable that varies with each sample. The table below provides an

overview of commonly used test statistics, depending on the presumed data distribution:

184
© 2014-2024 AnalystPrep.
 Hypothesis Test Test Statistic
Z-test Z- statistic (Normal distribution)
Chi-Square Test Chi-square statistic
t-test t-statistics
ANOVA F-statistic

We can subdivide the set of values that the test statistic can take into two regions: The non-

rejection region, which is consistent with the H 0, and the rejection region (critical region), which

is inconsistent with the H 0. If the test statistic has a value found within the critical region, we

reject the H 0.

As is the case with any other statistic, the distribution of the test statistic must be completely

specified under the H 0 when the H 0 is true.

The following is the list of test statistics and their distributions:

Test Test Statistic Test Statistic Number of

Subject Formula Distribution Degrees of
Freedom
X̄ −μ0
Single Mean t= s t − distribution n −1
√n
( X¯1 −X¯2 )−(μ1 −μ 2 )
Difference in t= t − distribution n1 + n2 − 1
2 2
s s
√ p+ p
n1 n2

Means
d̄−μ d0
Mean of t= t − distribution n −1
s d̄

Differences
s2 ( n −1)
Single χ2 = Chi-square n −1
σ2
0

Variance Distribution
S12
Difference in F= F-distribution n1 − 1,n 2 − 1
S22

variances
r√n −2
Correlation t= t-distribution n −2
√1−r2
2
( O ij−Eij)
Independence χ2 = ∑ m
i=1
Chi-square (r − 1) (c − 1)
Eij

(categorical data) Distribution

Where:

μ0 , μd 0 , and σ02 denote hypothesized values of the mean, mean difference, and variance in that

185
© 2014-2024 AnalystPrep.
order.

X̄ , b̄ s2 , s and r denote the sample mean of the differences, sample variance, sample standard

deviation, and correlation, in that order.

Oij and E ij are observed and expected frequencies, respectively, with r indicating the number of

rows and c indicating the number of columns in the contingency table.

Step 3: Specify the Level of Significance

The significance level represents the amount of sample proof needed to reject the null

hypothesis. First, let us look at type I and type II errors.

Type I and Type II Errors

When using sample statistics to draw conclusions about an entire population, the sample might

not accurately represent the population. This can result in statistical tests giving incorrect

results, leading to either erroneous rejection or acceptance of the null hypothesis. This

introduces the two errors discussed below.

Type I Error

Type I error occurs when we reject a true null hypothesis. For example, a type I error would

manifest in the rejection of H 0 = 0 when it is zero.

Type II Error

Type II error occurs when we fail to reject a false null hypothesis. In such a scenario, the

evidence the test provides is insufficient and, as such, cannot justify the rejection of the null

hypothesis when it is false.

Consider the following table:

186
© 2014-2024 AnalystPrep.
 Decision True Null Hypothesis False Null Hypothesis
(H0 ) (H 0 )
Fail to reject the Correct decision Type II error
null hypothesis
Reject null Type I error Correct decision
hypothesis

The level of significance, denoted by α, represents the probability of making a type I error, i.e.,

rejecting the null hypothesis when it is true. The confidence level complements the significance

level, (1 − α).

We use α to determine critical values that subdivide a distribution into the rejection and the non-

rejection regions. The figure below gives an example of the critical regions under a two-tailed

normal distribution and 5% significance level:

Consequently, β, the direct opposite of α, is the probability of making a type II error within the

bounds of statistical testing. The ideal but practically impossible statistical test would be one

that simultaneously minimizes α and β.

187
© 2014-2024 AnalystPrep.
The Power of a Test

The power of a test is the direct opposite of the significance level. The level of significance gives

us the probability of rejecting the null hypothesis when it is, in fact, true. On the other hand, the

power of a test gives us the probability of correctly discrediting and rejecting the null hypothesis

when it is false. In other words, it gives the likelihood of rejecting H 0 when, indeed, it is false.

Expressed mathematically,

Power of a test = 1 − β = 1 − P (type II error)

In a scenario with multiple test results for the same purpose, the test with the highest power is

considered the best.

Steps 4, 5, 6: State the Decision Rule, Calculate the Test

Statistic, and Make a Decision

The decision rule is the procedure that analysts and researchers follow when deciding whether

to reject or not reject a null hypothesis. We use the phrase "not to reject" because it's statistically

incorrect to "accept" a null hypothesis. Instead, we can only gather enough evidence to support

it.

Breaking Down the Decision Rule

The decision to reject or not reject a null hypothesis relies on the distribution of the test statistic.

The decision rule compares the calculated test statistic to the critical value.

If we reject the null hypothesis, the test is considered statistically significant. If not, we fail to

reject the null hypothesis, indicating insufficient evidence for rejection.

If a variable follows a normal distribution, we use the test's significance level to find critical

values corresponding to specific points on the standard normal distribution. These critical values

guide the decision-making process for rejecting or not rejecting a null hypothesis.

Before deciding whether to reject or not reject a null hypothesis, it's crucial to determine

188
© 2014-2024 AnalystPrep.
whether the test should be one-tailed or two-tailed. This choice depends on the nature of the

research question and the direction of the expected effect. Notably, the number of tails

determines the value of α (significance level). The following is a summary of the decision rules

under different scenarios.

Left One-tailed Test

H a : Parameter < X

Decision rule: Reject H 0 if the test statistic is less than the critical value. Otherwise, do not

reject H 0 .

Right One-tailed Test

H a : Parameter > X

Decision rule: Reject H 0 if the test statistic exceeds the critical value. Otherwise, do not reject

H 0.

189
© 2014-2024 AnalystPrep.
Two-tailed Test

H a : Parameter ≠ X (not equal to X)

Decision rule: Reject H0 if the test statistic is greater than the upper critical value or less than

the lower critical value.

190
© 2014-2024 AnalystPrep.
The p-Value in Hypothesis Testing

The p-value is the lowest level of significance at which we can reject a null hypothesis. The

probability of generating a test statistic would justify our rejection of a null hypothesis, assuming

that the null hypothesis is indeed true.

Breaking Down the p-value

When carrying out a statistical test with a fixed significance level (?) value, we merely compare

the observed test statistic with some critical value. For example, we might “reject an H 0 using a

5% test” or “reject an H 0 at a 1% significance level.” The problem with this ‘classical’ approach

is that it does not give us details about the strength of the evidence against the null

hypothesis.

Determination of the p-value gives statisticians a more informative approach to hypothesis

testing. The p-value is the lowest level at which we can reject an H 0. This means that the

strength of the evidence against an H0 increases as the p-value becomes smaller.

In one-tailed tests, the p-value is the probability below the calculated test statistic for left-tailed

tests or above the test statistic for right-tailed tests. For two-tailed tests, we find the probability

below the negative test statistic and add it to the probability above the positive test statistic. This

combines both tails for the p-value calculation.

Example: p-value

θ represents the probability of obtaining a head when a coin is tossed. Assume we tossed a coin

200 times, and the head came up in 85 out of the 200 trials. Test the following hypothesis at a

5% level of significance.

H 0 : θ = 0.5

H 1 : θ < 0.5

Solution

First, note that repeatedly tossing a coin follows a binomial distribution.

191
© 2014-2024 AnalystPrep.
Our p-value will be given by P (X < 85), where X follows a binomial (200,0.5), assuming the H0 is

true.

(85– 100)
= P ⌈Z < ⌉
√50
= P (Z < −2.12) = 1 − 0.9834 = 0.01660

(We have applied the Central Limit Theorem by taking the binomial distribution as approximately

normal.)

Since the probability is less than 0.05, the H0 is extremely unlikely, and we have strong evidence

against an H 0 that favors H1 . Therefore, clearly expressing this result, we could say:

“There is very strong evidence against the hypothesis that the coin is fair. We, therefore,

conclude that the coin is biased against heads.”

Remember, failure to reject a H0 does not mean it is true. It means there is insufficient evidence

to justify the rejection of the H0 given a certain level of significance.

192
© 2014-2024 AnalystPrep.
Question

A CFA candidate conducts a statistical test about the mean value of a random

variable X .

H 0 : μ = μ0 vs H1 : μ ≠ μ 0

She obtains a test statistic of 2.2. Given a 5% significance level, determine the p-

value.

A. 1.39%.

B. 2.78.

C. 2.78%.

Solution

The correct answer is C.

P − value = P(Z > 2.2) = 1 − P (Z < 2.2) = 1.39% × 2 = 2.78%

(We have multiplied by two since this is a two-tailed test.)

193
© 2014-2024 AnalystPrep.
Interpretation: The p-value (2.78%) is less than the significance level (5%). Therefore,

we have sufficient evidence to reject the H0 . In fact, the evidence is so strong that we

would also reject the H 0 at significance levels of 4% and 3%. However, at significance

levels of 2% or 1%, we would not reject the H0 since the p-value surpasses these

values.

194
© 2014-2024 AnalystPrep.
LOS 8b: Construct hypothesis tests and determine their statistical
significance, the associated Type I and Type II errors, and the power of
the test given a significance level

Hypothesis Test Concerning Single Mean

The z-test is the ideal hypothesis test when the sample's sampling distribution is normally

distributed or when the standard deviation is known.

The z-statistic is the test statistic used in hypothesis testing.

Testing H 0 : μ = μ0 Using the z-test

Given a random sample of size n from a normally distributed population with mean μ, variance

σ 2, and a sample mean X̄ , we can compute the z-statistic as follows:

(X̄ − μ0 )
z − statistic =
σ
( )
√n

Where:

X̄ is the sample mean.

?0 is the hypothesized mean of the population.

σ is the standard deviation of the population.

n is the sample size.

Once computed, the z-statistic is compared to the critical value corresponding to the test's

significance level. For example, if the significance level is 5%, the z-statistic is screened against

the upper or lower 95% point of the normal distribution (±1.96). The decision rule is to reject the

H0 if the z-statistic falls within the critical or rejection region.

Example: z-test

195
© 2014-2024 AnalystPrep.
Academics carried out a study on 50 former United States presidents and found an average IQ of

135. You are required to carry out a 5% statistical test to determine whether the average IQ of

presidents is greater than 130. (IQs are distributed normally, and previous studies indicate that

σ = 25.)

Solution

Step 1: State the hypothesis:

H 0 : μ ≤ 130

H 1 : μ > 130

Step 2: Identify the appropriate t-statistic:

(X̄ − μ0 )
z − statistic =
σ
( )
√n

( X̄−130)
Assuming the H 0 is true, ∼ N (0, 1)
( σ )
√n

Step 3: Specify the level of significance:

This is a right-tailed test. Therefore, we compare our test statistic to the upper 95% point of the

standard normal distribution (1.645).

Step 4: State the decision rule:

Reject the null hypothesis if the z-statistic is greater than 1.645.

Step 5: Calculate the test statistic:

(135– 130)
The z-statistic is = 1.414
( 25 )
√50

Step 6: Make a decision:

Since 1.414 is less than 1.645, we do not have sufficient evidence to reject the H 0 . As such, it

would be reasonable to conclude that the average IQ of U.S. presidents is not more than 130.

196
© 2014-2024 AnalystPrep.
The t-test

The t-test is based on the t-distribution. The test is appropriate for testing the value of a

population mean when:

σ is unknown.

The sample size is large (n ? 30), and if n < 30, the distribution must be normal or

approximately normal.

Testing H 0 : μ = μ0 Using the t-test

We compute a t-statistic with n − 1 degrees of freedom as follows:

(X̄ − μ 0 )
tn−1 =
s
( )
√n

Where:

X̄ is the sample mean.

μ0 is the hypothesized mean of the population.

s is the standard deviation of the sample.

n is the sample size.

Example: t-Test

Financial analysts in a certain equatorial country are interested in evaluating the potential

impact of rainfall on agricultural investments. They have gathered data on the annual rate of

rainfall (cm) over the last 10 years, as shown below:

{ 25, 26, 25, 27, 28, 29, 28, 27, 26, 25 }

Previously, the recorded average rainfall was 23 cm. The analysts want to find out if there's been

an increase in the average rainfall rate, which could impact agricultural investment. Conduct a

197
© 2014-2024 AnalystPrep.
statistical test at a 5% significance level to investigate this.

Solution

Follow the steps outlined above.

Step 1: State the hypothesis:

As always, you should begin by stating the hypothesis:

H 0 : μ ≤ 23

H 1 : μ > 23

Step 2: Identify the appropriate t-statistic:

If we assume that the annual rainfall quantities are distributed normally and recorded

independently, then:

(X̄ − μ0 )
∼ tn−1
s
( )
√n

Please, confirm that X̄ = 26.6 and s = 1.43

Step 3: Specify the level of significance:

α = 5% (right − tailed)

Step 4: State the decision rule:

Reject the null hypothesis if the t-statistic is greater than t0.05 ,9 = 1.833

Step 5: Calculate the test statistic:

(26.6 − 23)
Therefore, our t-statistic = = 7.96
( 1.43)
√10

Step 6: Make a decision:

Our test statistic (7.96) is greater than the upper 95% point of the t0.05 ,9 distribution (1.833).

198
© 2014-2024 AnalystPrep.
Therefore, we have sufficient evidence to reject the H0. As such, it is reasonable to conclude

that the average annual rainfall has increased from its former long-term average of 23.

199
© 2014-2024 AnalystPrep.
 Question

What is the value of t in the example above if the significance level is reduced from

5% to 0.5%, and does this change the decision rule?

A. 2.02; it does not change the decision rule.

B. 3.25; it does not change the decision rule.

C. 3.25; it changes the decision rule.

Solution

The correct answer is B.

A quick glance at the t0.005,9 distribution when α = 0.5% gives a value of 3.25.

However, the evidence against the H 0 is overwhelming since our test statistic (7.96)

is still greater than 3.25. As such, the conclusion would remain unchanged.

Hypothesis Test Concerning the Equality of the Population

Means

Analysts are often interested in establishing whether there exists a significant difference

between the means of two different populations. For instance, they might want to know whether

200
© 2014-2024 AnalystPrep.
the average returns for two subsidiaries of a given company exhibit a significant variance. Such

a test may then be used to make decisions regarding resource allocation or the reward of the

directors. Before embarking on such an exercise, ensuring that the samples taken are

independent and sourced from normally distributed populations is paramount. It can either be

assumed that the population variances are equal or unequal. In this reading, we will assume that

the population variances are equal.

Assume that μ 1 is the mean of the first population while μ2 is the mean of the second population.

In testing the equality of two population means, we wish to determine whether they are equal. As

such, the hypotheses can be any of the following:

I. Two-sided:

H0 : μ 1 − μ 2 = 0 vs. H a ?μ 1 − μ 2 ≠ 0,

This can be stated as follows:

H 0 : μ1 = μ 2 vs. H a?μ1 ≠ μ 2

II. One-sided (right):

H0 : μ 1 − μ 2 ≤ 0 vs. H a ?μ 1 − μ 2 > 0,

This can be stated as follows:

H 0 : μ1 ≤ μ 2 vs. H a?μ1 > μ 2

III. One-sided (Left):

H0 : μ 1 − μ 2 ≥ 0 vs. H a ?μ 1 − μ 2 < 0,

This can be stated as follows:

201
© 2014-2024 AnalystPrep.
 H 0 : μ1 ≥ μ 2 vs. H a?μ1 < μ 2

However, note that tests such as H 0 : μ1 − μ 2 = 3 vs. H a?μ1 − μ 2 ≠ 3 are valid. The

process is similar in both cases.

Hypothesis Test Concerning Differences between Means With

Independent Samples

When testing for the difference between two population means, we assume that the two

populations are distributed normally. Further, we assume that they have equal and unknown

variances. We always make use of the student's t-distribution where the test statistic is given by:

(X̄ 1 − X̄ 2 ) − (μ1 − μ2 ) (X̄ 1 − X̄ 2 ) − (μ 1 − μ2 )

t= =
s2
p s2
p s p (√ 1 + 1
)
√n + n1 n2
1 n2

Where s 2p is the pooled estimator of the common variance and is given by:

(n1 − 1) s21 + (n2 − 1) s22

s2p =
n 1 + n2 − 2

Also, the variables are defined as follows:

X̄ 1 = Mean of the first sample.

X̄ 2 = Mean of the second sample.

s21 = variance of the first sample.

s22 = variance of the second sample.

n1 = Sample size of the first sample.

n2 = Sample size of the second sample.

Example 1: Hypothesis Test Concerning the Equality of the Population Means

202
© 2014-2024 AnalystPrep.
Nutritionists want to establish whether obese patients on a new special diet have a lower weight

than the control group. After six weeks, the average weight of 10 patients (group A) on the

special diet is 75kg, while that of 10 more patients of the control group (B) is 72kg. Carry out a

5% test to determine if the patients on the special diet have a lower weight.

Additional information: ∑ A2 = 59520 and ∑B 2 = 56430

Solution

As is the norm, start by stating the hypothesis:

H 0 : μ1 − μ 2 = 0 Vs Ha ?μ 1 − μ2 ≠ 0,

We assume that the two samples have equal variances, are independent, and are normally

distributed. Then, under H 0 ,

B̄ − Ā
∼ tm+n −2
S√ m1 + 1
n

Note that the sample variance is given by:

2
∑X 2 − n X̄
2
s =
n− 1

So,

{59520 − (10 ∗ 75 2 )}
SA2 = = 363.33
9
{56430 − (10 ∗ 72 2 )}
SB2 = = 510
9

Therefore,

(9 × 363.33 + 9 × 510)
S2p = = 436.665
(10 + 10 − 2)

And
(75 − 72)
203
© 2014-2024 AnalystPrep.
 (75 − 72)
Test statistic = = 0.3210
1 1
{√439.665 × √ ( 10 + )}
10

Our test statistic (0.3210) is less than the upper 5% point (1.734) of the t-distribution with 18

degrees of freedom.

Therefore, we do not have sufficient evidence to reject the H 0 at 5% significance. As such, it is

reasonable to conclude that the special diet has the same effect on body weight as the placebo.

Note to candidates: You could choose to work with the p-value and determine P (t18 > 0.937)

and then establish whether this probability is less than 0.05. Working out the problem this way

would lead to the same conclusion as above.

Example 2: Hypothesis Test Concerning the Equality of the Population Means

204
© 2014-2024 AnalystPrep.
Suppose we replace ‘>’ with ‘≠’ in H1 in the example above, would the decision rule change?

Replacing ‘>’ with ‘≠’ in H1 would change the test from a one-tailed one to a two-tailed test. We

would compute the test statistic just as demonstrated above. However, we would have to divide

the significance level by two and compare the test statistic to the lower and upper 2.5% points of

the t18-distribution (±2.101).

Since our test statistic lies within these limits (non-rejection region), the decision rule would

remain unchanged.

Hypothesis Test Concerning Differences between Means With Dependent

Samples

There are some challenges when testing the difference between two population means using

205
© 2014-2024 AnalystPrep.
independent samples. Variability within each sample, caused by factors unrelated to the

research, can obscure the real difference of interest. Random variation within a sample might be

so substantial that it obscures the actual difference caused by the specific phenomenon the

analyst is studying.

When we want to test the differences between means with dependent samples, we use the paired

comparison test (test of the mean of the differences).

Paired Comparisons Test

Assume that we have observations for the random variables XA and XB and that the samples are

dependent.

Organize the observations in pairs and denote the differences between the two paired

observations by di . That is d i = x Ai − x Bi where xA i and x Bi are the ith pair of observations

1, 2 ,… , n.

Also, let μ d be the population mean difference and μd0 be the hypothesized value for the

population mean difference. At this point, we can state the hypotheses:

Two-sided: H 0 : μd = μd0 versus H a : μ d ≠ μd0

One-sided (right side): H 0 : μ d ≤ μd0 versus H a : μ d > μd0

One-sided (left side): H 0 : μd ≥ μd0 versus H a : μ d < μd0

Practically, μd0 = 0.

We are considering normally distributed populations with unknown population variances. As

such, we will use the t-distributed statistic given by:

d̄ − μ d0
t=
sd̄

Where:

206
© 2014-2024 AnalystPrep.
 1 n
d̄ = ∑ d
n i=1 i
sd
s d̄ = sd̄ = = standard error of the mean differences
√n
sd = standard deviation of the differences

Note that the degree of freedom is n − 1 where n is the number of the paired observations.

Example 2: Paired Comparison Test

An analyst aims to compare the performance of the BCD High Growth Index and the BCD

Investment Grade Index. They collect data for both indexes over 2,050 days and calculate the

means and standard deviations, as shown in the table below:

BCD High Growth BCD Investment Difference

Index (%) Grade Index (%) (%)
Mean return 0.0183 0.0161 −0.0022
Standard 0.2789 0.3298 0.3321
deviation

Using a 5% significance level, determine whether the mean of the differences is different from

zero.

Solution

Step 1: State the hypothesis:

H0 : μ d0 = 0 vs Ha : μd0 ≠ 0

Step 2: Identify the appropriate t-statistic:

We use the t-statistic, calculated by:

d̄ − μ d0
t=
sd̄

Step 3: Specify the level of significance:

α = 5% (two-tailed test)

207
© 2014-2024 AnalystPrep.
Step 4: State the decision rule:

The degrees of freedom amount to n − 1 = 2 , 050 − 1 = 2 , 049; thus, the critical values are

±1.960 . Therefore, we will reject the null hypothesis if the calculated t-statistic is less than -1.96

or greater than 1.96.

Step 5: Calculate the test statistic:

Note that from the table d̄ = −0.0022 and sd = 0.3321 so that the t-statistic is given:

d̄ − μd0 −0.0022 − 0
t= = = −0.30
s d̄ 0.3321
√2050

Step 6: Make a Decision:

-0.30 falls within the bounds of the critical values of ±1.960 . As such, there is insufficient

evidence to show that the mean of the differences in returns differs from zero.

Test Variance and Equality of Variance

208
© 2014-2024 AnalystPrep.
Testing of a Single Variance

A chi-square test helps determine if a hypothesized variance value matches the true population

variance. Unlike other distributions in the CFA® Program, the chi-square distribution is

asymmetrical. Yet, as degrees of freedom increase, it approaches a more normal distribution.

As a natural consequence, the chi-square distribution has no negative values and is bounded by

zero. A chi-square statistic with (n − 1) degrees of freedom is computed as:

(n − 1) S 2
χ2n −1 =
σ02

Where:

n = Sample size.

S 2 = Sample variance.

σ02 = Hypothesized population variance.

209
© 2014-2024 AnalystPrep.
Example: Chi-square Test

For the 15-year period between 1995 and 2010, ABC's monthly return had a standard deviation

of 5%. John Matthew, CFA, wishes to establish whether the standard deviation witnessed during

that period still adequately describes the long-term standard deviation of the company's return.

To achieve this end, he collects data on the monthly returns recorded between January 1, 2015,

and December 31, 2016, and computes a monthly standard deviation of 4%.

Carry out a 5% test to determine if the standard deviation computed in the latter period differs

from the 15-year value.

Solution

As is the norm, start by writing down the hypothesis.

H 0 : σ02 = 0.0025

H 1 : σ2 ≠ 0.0025

Since the latter period has 24 months, n = 24 , the test statistic is:

(24 − 1) 0.0016
χ 224−1 = = 14.72
0.0025

This is a two-tailed test. As such, we have to divide the significance level by two and screen our

test statistic against the lower and upper 2.5% points of χ223 .

Consulting the chi-square table, the test statistic (14.72) lies between the lower (11.689) and the

upper (38.076) 2.5% points of the chi-square distribution.

210
© 2014-2024 AnalystPrep.
Note that you will be given a simplified critical value table in the exam situation.

211
© 2014-2024 AnalystPrep.
Evidently, we have insufficient evidence to reject the H 0. It is, therefore, reasonable to conclude

that the latter standard deviation value is close enough to the 15-year value.

Test Concerning the Equality of the Variances

To test the equality concerning the variances, we use the F-test. Assume that we have 2

independent random samples of sizes n1 and n2 from N(μ 1 , σ12 ) and N(μ 2 , σ22 ).

Also, let us consider a scenario where we have the sample variances as S12 and S22. The basic

situation usually involves the following hypothesis:

H 0 : σ12 = σ22

H a : σ12 ≠ σ22

S 12
The test statistic is ∼ Fn 1–1,n 2–1 under H 0.
2
S2

212
© 2014-2024 AnalystPrep.
The decision rule is to reject the null hypothesis if the test statistic falls within the critical region

of the F-distribution.

Example: F-test

An analyst is studying whether the population variance of returns on a commodity index changed

after the introduction of new trading guidelines. The first 320 weeks elapsed before the

guidelines were introduced, and the second 320 weeks came after the introduction. The analyst

gathers the data in the table below for 320 weeks of returns before and after the change in

guidelines.

Mean Weekly Variance of

Return (%) Returns
Before guidelines change 0.180 3.520
After guidelines change 0.090 2.967

Do the variances of returns differ before and after the guideline change? Employ a 5 percent

significance level.

213
© 2014-2024 AnalystPrep.
Solution

Step 1: State the hypothesis:

2
H 0 : σBefore 2
= σAfter
2 vs Ha : σBefore ≠ σAfter
2

Step 2: Identify the appropriate t-statistic:

s2Before
t=
s2After

Step 3: Specify the level of significance:

α = 5% (two-tailed)

Step 4: State the decision rule:

Left side = 0.803

Right side = 1.246

Reject the null if the calculated t-statistic is less than 0.803 and reject the null if the calculated t-

statistic is greater than 1.246.

Step 5: Calculate the test statistic:

s2Before 3.520
t= = = 1.1864
s2After 2.967

Step 6: Make a decision:

Fail to reject the null hypothesis because 1.1864 falls within the bounds of the critical values of

[0.80,1.246]. There is insufficient evidence to indicate that the weekly variances of returns are

different in the periods before and after the guidelines change.

214
© 2014-2024 AnalystPrep.
LOS 8c: Compare and contrast parametric and nonparametric tests, and
describe situations where each is the more appropriate type of test

Parametric Tests

Parametric tests are statistical tests in which we make assumptions regarding population

distribution. Such tests involve the estimation of the key parameters of a distribution. For

example, we may wish to estimate the mean or compare population proportions.

When conducting statistical tests, the choice of distribution directly influences how the test

statistic is calculated. The tests we've discussed are considered parametric tests. For example,

assuming a parameter follows a normal distribution leads to the computation of the z-statistic.

During parametric testing, approximating normal distributions for non-normal data may be

required. This approximation is valid due to the central limit theorem, which states that as

sample sizes increase, non-normal distributions tend to become more normal.

Parametric tests are generally considered to be stronger than nonparametric ones.

Nonparametric Tests

Nonparametric tests, sometimes known as distribution-free tests, don't assume anything about

the parameter's distribution being studied. Researchers turn to nonparametric testing when they

have concerns about factors other than the parameter's distribution.

The following table gives the alternative nonparametric tests for the parametric tests.

Parametric Test Non-parametric Test

Test concerning t-distributed test Wilcoxon signed-rank test
single mean
Tests concerning z-distributed test Mann–Whitney U test
differences between (Wilcoxon rank sum test)
means
Test concerning mean t-distributed test Wilcoxon signed-rank
differences (pair test Sign test
comparison test)

215
© 2014-2024 AnalystPrep.
Situations Where Nonparametric Tests are Appropriate

The data do not meet distributional assumptions:

This happens when the distributional assumptions of the parametric tests are not met. For

instance, we may find parametric tests such as t-test are inappropriate because the sample size

is small and may be drawn from non-normally distributed. As such, a nonparametric test is

appropriate.

When there are outliers:

Outliers can affect the parametric statistics. On the other hand, outliers do not affect parametric

tests.

Consider a situation where we want to establish the center of a rather skewed distribution, such

as that of the income of the residents of a given city. While the majority of the residents could be

categorized as the middle class, the presence of just a few billionaires in a sample can greatly

increase the mean income. Such a mean, therefore, may not provide a very reliable or realistic

measure of income.

Instead, it may be more appropriate to use the median. Compared to the mean, the median can

better represent the center of the income distribution. This is due to the fact that 50% of the

residents will be above the median and the remaining 50% below it.

In summary, “outliers” affect the mean when dealing with skewed data. The median, on the other

hand, sticks closer to the center of the distribution.

When data is given in the form of ranks or uses an ordinal scale:

Although nonparametric tests are usually easier to conduct than parametric ones, they do not

have as much statistical power. Nonetheless, they provide an efficient tool for analyzing ordinal,

ranked, or very skewed data.

When the hypotheses do not concern a parameter:

216
© 2014-2024 AnalystPrep.
We often use nonparametric tests, such as the runs test, when our goal is to determine if a

sample from a population isn't random. Since randomness isn't a parameter, nonparametric tests

are the right choice in such cases.

Nonparametric inference:

Nonparametric methods make our statistical analysis broader. They work with limited

assumptions and can be used for ordered data. Plus, they handle questions that aren't tied to

specific parameters.

Nonparametric tests are commonly used alongside parametric tests. They help analysts

understand how sensitive the statistical results are to the assumptions of parametric tests. But

when the conditions for a parametric test are met, we usually choose it over nonparametric tests.

We prefer parametric tests because they often have more statistical power, which means they are

better at detecting when the null hypothesis is incorrect.

217
© 2014-2024 AnalystPrep.
 Learning Module 9: Parametric and Non Parametric Tests of
Independence

LOS 9a: explain parametric and non-parametric tests of the hypothesis

that the population correlation coefficient equals zero and determine
whether the hypothesis is rejected at a given level of significance

Parametric versus Non-parametric Tests of Independence

A parametric test is a hypothesis test concerning a population parameter used when the data has

specific distribution assumptions. If these assumptions are not met, non-parametric tests

are used.

In summary, researchers use non-parametric testing when:

Data do not meet distributional assumptions.

There are outliers.

Data is given in the form of ranks.

The hypothesis test objective does not concern a parameter.

Hypotheses Concerning Population Correlation Coefficient

We frequently compare the population correlation coefficient to zero when testing for

correlation. This helps us determine whether there's a relationship between the variables. The

population correlation coefficient, represented by ρ, is used to test the relationship. There are

three possible hypotheses:

Two-sided; H0 : ρ = 0 versus H a : ρ ≠ 0.

One-sided right side; H 0 : ρ ≤ 0 versus H a : ρ > 0 .

One-sided left side; H 0 : ρ ≥ 0 versus H a : ρ < 0 .

Let's assume that we have variables X and Y. The sample correlation, rXY , tests the above

218
© 2014-2024 AnalystPrep.
hypotheses.

Parametric Test of a Correlation

The parametric pairwise correlation coefficient, also known as Pearson correlation, is

used to test the correlation in a parametric test. The formula for the sample correlation involves

the sample covariance between the X and Y variables and their respective standard deviations,

which is expressed as:

SXY
r=
SX SY

Where:

SXY = Sample covariance between the X and Y variables.

SX = Standard deviation of the X variable.

SY = Standard deviation of the Y variable.

A t-test can determine if the null hypothesis should be rejected using the sample correlation, r if

the two variables are normally distributed. The formula for the t-test is:

r√n − 2
t=
√(1 − r2 )

Where:

r= Sample correlation.

n= Sample size.

(n − 2)= Degrees of freedom.

The test statistic follows a t-distribution with n − 2 degrees of freedom. From the equation above,

it is easy to see that the sample size, n, increases, and the degrees of freedom increase. In other

words, as the sample size n increases, the power of the test increases. This implies that a false

219
© 2014-2024 AnalystPrep.
null hypothesis will likely be rejected as the sample size increases.

Example: Parametric Test of a Correlation

The table below shows the sample correlations between the monthly returns of five different

sector-specific exchange-traded funds (ETFs) and the overall market index (Market 1). There are

48 monthly observations, and the following ETFs are included in the analysis:

ETF 1 ETF 2 ETF 3 ETF 4 ETF 5 Market 1

ETF 1 1
ETF 2 0.8214 1
ETF 3 0.5672 0.6438 1
ETF 4 0.4276 0.5789 0.4123 1
ETF 5 0.7121 0.7942 0.6896 0.5614 1
Market 1 0.8375 0.9096 0.7223 0.6954 0.7919 1

Using a 1% significance level and the following hypotheses: H0 : ρ = 0 versus Ha : ρ ≠ 0, calculate

the t-statistic for the correlation between ETF 2 and ETF 4. Based on the calculated t-statistic,

draw a conclusion about the significance of the correlation using the following sample t-table:

df p = 0.10 p = 0.05 p = 0. 025 p = 0.01 p = 0. 005

31 1.309 1.696 2.040 2.453 2.744
32 1.309 1.694 2.037 2.449 2.738
33 1.308 1.692 2.035 2.445 2.733
34 1.307 1.691 2.032 2.441 2.728
35 1.306 1.690 2.030 2.438 2.724
36 1.306 1.688 2.028 2.434 2.719
37 1.305 1.687 2.026 2.431 2.715
38 1.304 1.686 2.024 2.429 2.712
39 1.304 1.685 2.023 2.426 2.708
40 1.303 1.684 2.021 2.423 2.704
41 1.303 1.683 2.020 2.421 2.701
42 1.302 1.682 2.018 2.418 2.698
43 1.302 1.681 2.017 2.416 2.695
44 1.301 1.680 2.015 2.414 2.692
45 1.301 1.679 2.014 2.412 2.690
46 1.300 1.679 2.013 2.410 2.687
47 1.300 1.678 2.012 2.408 2.685
48 1.299 1.677 2.011 2.407 2.682

Solution

220
© 2014-2024 AnalystPrep.
To test the significance of the correlation between ETF 2 and ETF 4, we will use the t-test

formula:

r√n − 2
t=
√(1 − r2 )

Where:

r = Sample correlation coefficient (in this case, rEF T2 ,ET F 4 = 0.5789 ).

n = Number of observations (48 in this case).

Now, let's calculate the t-statistic:

r√n − 2 0.5789√48 − 2
t= = = 4.815
√(1 − r2 ) √1 − 0.57892

The calculated t-statistic for the correlation between ETF2 and ETF4 is 4.815.

At the 1% significance level, with a two-tailed test and degrees of freedom,

df = n − 2 = 46, the critical t-value is approximately ±2.687.

Conclusion: We reject the null hypothesis since our calculated t-statistic (4.815) is greater than

the critical value (+2.687). This indicates sufficient evidence to suggest that the correlation

between ETF 2 and ETF 4 significantly differs from zero.

Non-Parametric Test of Correlation: The Spearman Rank

Correlation Coefficient

The Spearman rank correlation coefficient, rS , is a non-parametric test used to examine the

relationship between two data sets when the population deviates from normality.

The Spearman rank correlation coefficient is like the Pearson correlation coefficient. The

difference is that the Spearman coefficient is calculated based on the ranks of variables in the

samples.

221
© 2014-2024 AnalystPrep.
Consider two variables, X and Y . We need to calculate Spearman's Rank Correlation rS.

Steps of Calculating Spearman's Rank Correlation Coefficient, rS

Rank the observations of each variable X and Y in descending order. Note that when

there are tied values in the data, their ranks are calculated by taking the average of the

ranks that would have been assigned to those values if they were not tied.

Find the difference between the ranks for each pair of observations.

Square the difference and calculate the sum of the difference, that is, ∑ di .

Use the following formula to find rS:

6 ∑ni=1 d i2
rs = 1 −
n (n 2 − 1)

Where; di =The difference between the ranks for each pair of observationsn= Sample

size.

Example: Calculating Spearman's Rank Correlation Coefficient

An analyst is studying the relationship between returns for two sectors, steel and cement, over

the past 5 years using Spearman's rank correlation coefficient. The hypotheses are H 0 : rS = 0

and H a : rS ≠ 0 . The returns of both sectors are provided below.

Year Steel sector returns Cement sector returns

1 10% 8%
2 6% 7%
3 9% 5%
4 12% 6%
5 8% 9%

The Spearman's rank correlation coefficient is closest to:

Solution

222
© 2014-2024 AnalystPrep.
 Year Steel Cement Rank Rank D d2
sector sector order order
returns returns for X for Y
(X) (Y)
1 10% 8% 2 2 0 0
2 6% 7% 5 3 2 4
3 9% 5% 3 5 −2 4
4 12% 6% 1 4 −3 9
5 8% 9% 4 1 3 9
Sum = 26

We can now use the formula:

6 ∑ ni=1 d i2 (6 × 26)
rs = 1 − = 1− [ ] = 1 − 1.3
n (n 2 − 1) 5 × (5 2 − 1)
rs = −0.3

This indicates a very weak negative correlation between the returns of the steel and cement

sectors.

Hypothesis Test for the Spearman Rank Correlation

The hypothesis test on the Spearman Rank depends on the sample size. If the sample size is

small (n ≤ 30), we would need a specialized table of critical value. On the other hand, if the

sample size is large (n > 30), we can perform a t-test using the test statistic similar to that of

Pearson correlation:

rs √n − 2
t=
√(1 − r2s )

Consider the above example. Assume we want to conduct a hypothesis test at a 5% significance

level. The hypotheses statement is H 0 : rS = 0 and H a : rS ≠ 0

223
© 2014-2024 AnalystPrep.
Question

Assume an investment analyst, John Smith, is studying the relationship between two

stocks, X and Y . Based on 100 observations, he has found that SXY = 10, SX = 2, and

SY = 8. Smith needs to find the sample correlation rXY and use it to perform a t-test

to determine if there is a significant correlation between the returns of stocks X and

Y . The critical value for the test statistic at the 0.05 level of significance is

approximately 1.96. He should conclude that the statistical relationship between X

and Y is:

A. Significant because the test statistic falls outside the range of the critical

values.

B. Significant, because the absolute value of the test statistic is less than the

critical value.

C. Insignificant because the test statistic falls outside the range of the critical

values.

Solution

The correct answer is A.

Note that the sample correlation coefficient, rXY is calculated using the following

formula:

SXY
rXY =
SX SY

Substituting the given values in this formula, we get:

10
rXY = = 0.625
2× 8

To test the significance of the sample correlation, we can use a t-test with the

following null and alternative hypotheses: H 0 = ρ = 0 and H ∝ = ρ ≠ 0

224
© 2014-2024 AnalystPrep.
The test statistic for this test is calculated using the following formula:

r√ n − 2
t=
√1 − r2

Where:

r = Sample correlation coefficient.

n = The Number of observations.

Substituting the given values into this formula, we get:

0.625√ 100 − 2 6.1872

t= ≈ = 7.9262
√1 − 0.625 2 0.7806

The critical value for the test statistic at the 0.05 level of significance is

approximately 1.96.

Since our calculated test statistic (7.9262) is greater than the upper bound of the

critical values for the test statistic (1.96), we reject the null hypothesis. This indicates

sufficient evidence to suggest that the correlation between X and Y is significantly

different from zero.

Therefore, John Smith should conclude that the statistical relationship between X and

Y is significant because the test statistic falls outside the range of the critical values

(Option A).

225
© 2014-2024 AnalystPrep.
LOS 9b: Explain tests of independence based on contingency table data

With categorical or discrete data, correlation is not suitable for assessing relationships between

variables. Instead, we use a non-parametric test called the chi-square test of independence,

which employs a chi-square distributed test statistic.

We employ a contingency table to structure the data when examining the connection between

two categorical variables. Subsequently, we apply a test of independence utilizing a chi-square

distribution to assess whether a noteworthy relationship exists between these variables. The test

statistic is calculated as follows:

2
m (Oij − Eij )
χ=∑
i=1 (E ij )

Where:

(Total row i)×(Total column j)

Ei j = Overall Total

m= Number of cells in the table, the Number of groups in the first class, multiplied by the

number of groups in the second class.

Oij = Number of observations in each cell of row i and column j (i.e., observed frequency).

Ei j= Expected number of observations in each cell of row i and column j, assuming independence

(i.e., expected frequency).

The degrees of freedom are given by:

Degrees of freedom = (r − 1)(c − 1)

Where:

r= Number of rows.

c= Number of columns.

Example: Testing Independence Based on Contingency Table Data

226
© 2014-2024 AnalystPrep.
The following contingency table shows the responses of two categories of investors (employed

vs. retired) with regard to their primary investment objectives (growth, income, or both). The

total sample size is 173.

Growth Income Both Total

Employed 52 25 10 87
Retired 32 47 7 86
Total 84 72 17 173

Use a 95% significance level to test whether there is any significant difference between

employed and retired investors concerning primary investment objectives.

Solution

H 0: There is no significant difference between employed and retired investors with regard to

primary investment objectives.

H α: There is a significant difference between employed and retired investors with regard to

primary investment objectives.

Step 1: We calculate the expected frequency of investors by their category (employed vs.

retired) and investment objective using the following formula:

(Total row i) × (Total column j)

Ei j =
Overall Total

Growth Income Both Total

(87×84) (87×72) (87×17)
Employed = 42.24 = 36.20 = 8.55 87
173 173 173
(86×84) (86×72) (86×17)
Retired = 41.75 = 35.79 = 8.45 86
173 173 173
Total 84 72 17 173

Step 2: We calculate the scaled squared deviation for each combination of investor category and

investment objective as follows:

227
© 2014-2024 AnalystPrep.
 Growth Income Both
(52−42) 2 (25−36) 2 (10−9)2
Employed 42
= 2.254 36
= 0.469 9
= 0.246
(32−42) 2 (47×36) 2 (7−8) 2
Retired 42
= 2.280 36
= 3.510 8
= 0.349
Total 4.534 6.979 0.495

Step 3: We calculate the value of χ2 :

χ2 = 4.534 + 6.979 + 0.495 = 12.008

Step 4: The critical value of X 2 is 5.99. It is determined as follows:

There are (r − 1)(c − 1) = (2 − 1) × (3 − 1) = 2 degrees of freedom.

It is a one-sided test with a 5% level of significance.

228
© 2014-2024 AnalystPrep.
Decision rule: The calculated value of χ2 = 12.008 is greater than the critical value of 5.99. As

such, sufficient evidence supports the conclusion that retired and employed investors have

different primary investment objectives.

229
© 2014-2024 AnalystPrep.
Question

Regarding the chi-square test of independence, which statement is accurate? The chi-

square test of independence is:

A. A parametric hypothesis test.

B. Used to test whether two categorical variables are related to each other.

C. Used to test whether two continuous variables are related to each other.

Solution

The correct answer is B. The chi-square test of independence is a non-parametric

hypothesis test that can be used to test whether two categorical variables are related.

A is incorrect because the chi-square test of independence is non-parametric, not

parametric.

C is incorrect because the chi-square test of independence is used for categorical

variables, not continuous variables.

230
© 2014-2024 AnalystPrep.
 Learning Module 10: Simple Linear Regression

LOS 10a: describe a simple linear regression model, how the least
squares criterion is used to estimate regression coefficients, and the
interpretation of these coefficients

Linear regression is a mathematical method used for analyzing how the variation in one variable

can explain the variation in another variable.

Let Y be the variable we wish to explain. As such, the observation of this variable is Yi , and Y¯ is

the mean of the sample size n. The variation of Y is given by:

n
2
Variation of Y = ∑ (Y i − Y¯ )
i=1

Our main objective is to explain what causes this variation, usually called the sum of squares

total (SST).

By definition of the regression, we need to explain the variation of Y with another variable. Let X

be the explanatory variable. As such, the observations of X will be denoted by Xi and X̄ sample

mean of size n . The variation of X is given by:

n
2
Variation of X = ∑ (Xi − X̄ )
i=1

To visualize the relationship between variables X and Y, you can use a scatter plot, also known as

a scattergram. In this type of plot, the variable you want to explain (Y) is usually plotted on the

vertical axis. In contrast, the explanatory variable (X) is placed on the horizontal axis to show the

relationship between their variations.

For example, consider the following table. We wish to use linear regression analysis to forecast

inflation, given unemployment data from 2011 to 2020.

231
© 2014-2024 AnalystPrep.
 Year Unemployment Rate Inflation Rate
2011 6.1% 1.7%
2012 7.4% 1.2%
2013 6.2% 1.3%
2014 6.2% 1.3%
2015 5.7% 1.4%
2016 5.0% 1.8%
2017 4.2% 3.3%
2018 4.2% 3.1%
2019 4.0% 4.7%
2020 3.9% 3.6%

In this scenario, the Y variable is the inflation rate, and the X axis is the unemployment rate. A

scatter plot of the inflation rates against unemployment rates from 2011 to 2020 is shown in the

following figure.

Dependent and Independent Variables

A dependent variable, often denoted as YYY, is the variable we want to explain. In contrast, an

independent variable, typically denoted as XXX, explains variations in the dependent variable.

232
© 2014-2024 AnalystPrep.
The independent variable is also referred to as the exogenous, explanatory, or predicting

variable.

In our example above, the inflation rate is the dependent variable, and the unemployment rate is

the independent variable.

To understand the relationship between dependent and independent variables, we estimate a

linear relationship, usually a straight line. When there's one independent variable, we use simple

linear regression. If there are multiple independent variables, we use multiple regression.

This reading focuses on linear regression.

Least Squares Criterion

In simple linear regression, we assume linear relationships exist between the dependent and

independent variables. The aim is to fit a line to the observations of X (Xi s) and Y (Y is) to

minimize the squared deviations from the line. To accomplish this, we use the least squares

criterion.

The following is a simple linear regression equation:

Y = b 0 + b 1 X1 + εi, i = 1, 2, … , n

Where:

Y = Dependent variable.

b0 = Intercept.

b1 = Slope coefficient.

X = Independent variable.

ε = Error term (Noise).

b0 and b 1 are known as regression coefficients. The equation above implies that the dependent

233
© 2014-2024 AnalystPrep.
is equivalent to the intercept (b0 ) plus the product of the slope coefficient (b 1 ) and the

independent variable plus the error term.

The error term is equal to the difference between the observed value of Y and the one expected

from the underlying population relation between X and Y

Below is an illustration of a simple linear regression model.

As stated earlier, linear regression calculates a line that best fits the observations. In the

following image, the line that best fits the regression is clearly the blue one:

234
© 2014-2024 AnalystPrep.
Note that we cannot directly observe the population parameters b0 and b 1. As such, we observe

their estimates, ^
b 0 and ^
b 1 . They are the estimated parameters of the population using a sample.

In simple linear regression, ^

b 0 and ^
b 1 are such that the sum of squared vertical distances is

minimized.

Specifically, we concentrate on the sum of the squared differences between observations Yi and

the respective estimated value Y^i on the regression line, also called the sum of squares error

(SSE).

Note that,

Y^i = ^
b 0 +^
b 1 Xi + e2i

As such,

235
© 2014-2024 AnalystPrep.
 n 2 2 n n
SSE = ∑ (Y i − ^
b 0 −^
b 1 Xi) = ∑ (Yi − Y^i ) = ∑ e2i
i =1 i =1 i=1

Note that the residual for the ith observation (ei = Y i − Y^i ) is different from the error term (εi ).

The error term is based on the underlying population, while the residual term results from

regression analysis on a sample.

Conventionally, the sum of the residuals is zero. As such, the aim is to fit the regression line in a

simple linear regression that minimizes the sum of squared residual terms.

Estimation and Interpretation of Regression Coefficients

The Slope Coefficient β^1

For a simple linear regression, the slope coefficient is estimated as the ratio of the Cov(X, Y ) and

V ar(X):

∑n=1 (Y i−Y¯ )(Xi−X̄ )

i
Cov (X, Y ) n −1 ∑ni=1 (Y i − Y¯) (X i − X̄)
^
b1= = =
V ar (X) ∑n=1 (Xi−X̄ )2 2
i ∑ni=1 (Xi − X̄)
n −1

The slope coefficient is defined as the change in the dependent variable caused by a one-unit

change in the value of the independent variable.

The Intercept β^0

The intercept is estimated using the mean of X and Y as follows:

^
b 0 = Y¯ − ^
b 1 X̄

Where:

Y^ = Mean of Y .

X^ = Mean of X.

236
© 2014-2024 AnalystPrep.
The intercept is the estimated value of the dependent variable when the independent variable is

zero. The fitted regression line passes through the point equivalent to the means of the

dependent and the independent variables in a linear regression model.

Example: Estimating Regression Line

Let us consider the following table. We wish to estimate a regression line to forecast inflation,

given unemployment data from 2011 to 2020.

Year Unemployment Rate% (X is) Inflation Rate% (Y i s)

2011 6.1 1.7
2012 7.4 1.2
2013 6.2 1.3
2014 6.2 1.3
2015 5.7 1.4
2016 5.0 1.8
2017 4.2 3.3
2018 4.2 3.1
2019 4.0 4.7
2020 3.9 3.6

We can create the following table:

2 2
Year Unemployment Inflation (Y i − Y¯ ) (Xi − X̄) (Yi − Y¯)
Rate% (Xi s) Rate% (Y i s) (Xi − X̄ )
2011 6.1 1.7 0.410 0.656 −0.518
2012 7.4 1.2 1.300 4.452 −2.405
2013 6.2 1.3 1.082 0.828 −0.946
2014 6.2 1.3 1.082 0.828 −0.946
2015 5.7 1.4 0.884 0.168 −0.385
2016 5.0 1.8 0.292 0.084 0.157
2017 4.2 3.3 0.922 1.188 −1.046
2018 4.2 3.1 0.578 1.188 −0.828
2019 4.0 4.7 5.570 1.664 −3.044
2020 3.9 3.6 1.588 1.932 −1.751
Sum 52.90 23.4 13.704 12.989 −11.716
Arithmetic 5.29 2.34
Mean

From the table above, we estimate the regression coefficients:

Cov (X Y ) ∑n (Y − Y¯ ) (X − X̄ )
237
© 2014-2024 AnalystPrep.
 Cov (X , Y ) ∑ni=1 (Y i − Y¯ ) (Xi − X̄ ) −11.716
^
b1 = = = = −0.9020
V ar (X) ∑ni=1 (Xi − X̄ )
2 12.989
^
b 0 = Y¯ − ^
b 1X̄ = 2.34 − (−0.9020) × 5.29 = 7.112

As such, the regression model is given by:

Y^ = 7.112 − 0.9020Xi + εi

From the above regression model, we can note the following:

The inflation rate is 7.112% if the unemployment rate is 0% (theoretically speaking).

If the unemployment rate increases (decreases) by one unit, say, from 2% to 3%–the

inflation rate decreases(increases) by 0.9020%.

In general,

If the slope is positive, a unit increase(decrease) in the independent variable results in

an increase(decrease) in the dependent variable.

If the slope is negative, a one-unit increase(decrease) in the independent variable

results in a decrease(increase) in the dependent variable.

Furthermore, with the estimated regression model, we can predict the values of the dependent

variable based on the value of the independent variable. For instance, if the unemployment rate

is 4.5%, then the predicted value of the dependent variable is:

Y^ = 7.112 − 0.9020 × 4.5 = 3.05%

In practice, analysts perform regression analysis using statistical functions in software like

Excel, statistical tools like R, or programming languages such as Python.

Cross-sectional and Time Series Regressions

Regression analysis is commonly used with cross-sectional and time series data. In cross-

238
© 2014-2024 AnalystPrep.
sectional analysis, you compare X and Y observations from different entities, like various

companies in the same time period. For instance, you might analyze the link between a

company's R&D spending and stock returns across multiple firms in a year.

Time-series regression analysis involves using data from various time periods for the same entity,

like a company or an asset class. For instance, an analyst might examine how a company's

quarterly dividend payouts relate to its stock price over multiple years.

239
© 2014-2024 AnalystPrep.
Question

The independent variable in a regression model is most likely the:

A. Predicted variable.

B. Predicting variable.

C. Endogenous variable.

Solution

The correct answer is B.

An independent variable explains the variation of the dependent variable. It is also

called the explanatory variable, exogenous variable, or the predicting variable.

A and C are incorrect. A dependent variable is a variable predicted by the

independent variable. It is also known as the predicted variable, explained variable,

or endogenous variable.

240
© 2014-2024 AnalystPrep.
LOS 10b: explain the assumptions underlying the simple linear
regression model, and describe how residuals and residual plots indicate
if these assumptions may have been violated

Assume that we have samples of size n for dependent variable Y and independent variable X . We

wish to estimate the simple regression of Y and X. The classic normal linear regression model

assumptions are as follows:

Linearity: A linear relationship implies that the change in Y due to a one-unit change in X is

constant, regardless of the value X takes. If the relationship between the two is not linear, the

regression model will not accurately capture the trend, resulting in inaccurate predictions. The

model will be biased and underestimate or overestimate Y at various points. For example, the

model Y = b 0 + b1 eb1x is nonlinear in b1 . For this reason, we should not attempt to fit a linear

model between X and Y . It also follows that the independent variable, X , must be non-stochastic

(must not be random). A random independent variable rules out a linear relationship between

the dependent and independent [Link] addition, linearity means the residuals should not

exhibit an observable pattern when plotted against the independent variable. Instead, they

should be completely random. In the example below, we're looking at a scenario where the

residuals appear to show a pattern when plotted against the independent variable, X . This

effectively indicates a nonlinear relation.

241
© 2014-2024 AnalystPrep.
 Normality Assumption: This assumption implies that the error terms (residuals) must follow

a normal distribution. It's important to note that this doesn't mean the dependent and

independent variables must be normally distributed. However, it's crucial to check the

distribution of the dependent and independent variables to identify any outliers. A histogram of

the residuals can be used to detect if the error term is normally distributed. A symmetric bell-

shaped histogram indicates that the normality assumption is likely to be true.

Homoskedasticity: Homoskedasticity implies that the variance of the error terms is constant

across all observations. Mathematically, this is expressed as:

E (ϵ 2i ) = σϵ2, i = 1 , 2 , … , n

If the variance of residuals varies across observations, then we refer to this as heteroskedasticity

242
© 2014-2024 AnalystPrep.
(not homoscedasticity). We plot the least square residuals against the independent variable to

test for heteroscedasticity. If there is an evident pattern in the plot, that is a manifestation of

heteroskedasticity.

In case residuals and the predicted values increase simultaneously, then such a situation is

known as heteroscedasticity (or heteroskedasticity).

243
© 2014-2024 AnalystPrep.
 Independence Assumption: The independence assumption implies that the observations Xi

and Y i are independent of each other. Failure to satisfy this assumption implies the variables are

not independent, and thus, residuals will be correlated. To ascertain this assumption, we visually

and statistically analyze the residuals to check whether residual shows exhibit a pattern.

244
© 2014-2024 AnalystPrep.
Question

A regression model with one independent variable requires several assumptions for

valid conclusions. Which of the following statements most likely violates those

assumptions?

A. The independent variable is random.

B. The error term is distributed normally.

C. There exists a linear relationship between the dependent variable and the

independent variable.

Solution

The correct answer is A.

Linear regression assumes that the independent variable, X, is NOT random. This

ensures that the model produces the correct estimates of the regression coefficients.

B is incorrect. The assumption that the error term is distributed normally allows us

to easily test a particular hypothesis about a linear regression model.

C is incorrect. Essentially, the assumption that the dependent and independent

variables have a linear relationship is the key to a valid linear regression. If the

parameters of the dependent and independent variables are not linear, then the

estimation of that relation can yield invalid results.

245
© 2014-2024 AnalystPrep.
LOS 10c: calculate and interpret measures of fit and formulate and
evaluate tests of fit and of regression coefficients in a simple linear
regression

The sum of Squares Total (SST) and Its Components

The sum of Squares Total (total variation) is a measure of the total variation of the dependent

variable. It is the sum of the squared differences of the actual y-value and mean of y-

observations.

n
2
S ST = ∑ (Y i − Y¯ )
i=1

The Sum of Squares Total contains two parts:

i. The Sum of Square Regression (SSR).

ii. The sum of Squares Error (SSE).

i. The sum of Squares Regression (SSR): The sum of squares regression measures the

explained variation in the dependent variable. It is given by the sum of the squared

differences of the predicted y-value Y^i , and mean of y-observations, Y¯ :

n 2
S SR = ∑ (Y^i − Y¯ )
i=1

ii. The Sum of Squared Errors (SSE): The sum of squared errors is also called the

residual sum of squares. It is defined as the variation of the dependent variable

unexplained by the independent variable. SSE is given by the sum of the squared

differences of the actual y-value (Y i ) and the predicted y-values, Y^i .

n 2
SSE = ∑ (Y i − Y^i )
i=1

Therefore, the sum of squares total is given by:

246
© 2014-2024 AnalystPrep.
 Sum of Squares Total = Explained Variation + Unexplained Variation
= SSR + SSE

The components of the total variation are shown in the following figure.

For example, consider the following table. We wish to use linear regression analysis to forecast

inflation, given unemployment data from 2011 to 2020.

247
© 2014-2024 AnalystPrep.
 Year Unemployment Rate (%) Inflation Rate (%)
2011 6.1 1.7
2012 7.4 1.2
2013 6.2 1.3
2014 6.2 1.3
2015 5.7 1.4
2016 5.0 1.8
2017 4.2 3.3
2018 4.2 3.1
2019 4.0 4.7
2020 3.9 3.6

Remember that we had estimated the regression line to be Y^ = 7.112 − 0.9020Xi + εi . As such,

we can create the following table:

Year Unemployment Inflation Predicted Variation Variation Variation

Rate %(Xi ) Rate % Unemployment to be Unexplained Explained
(Y i) ^
rate (Y i ) Explained.
2 2 2
(Y i − Y¯) (Yi − Y^i ) (Y^i − Y¯ )
2011 6.1 1.7 1.610 0.410 0.008 0.533
2012 7.4 1.2 0.437 1.300 0.582 3.621
2013 6.2 1.3 1.520 1.082 0.048 0.673
2014 6.2 1.3 1.520 1.082 0.048 0.673
2015 5.7 1.4 1.971 0.884 0.326 0.136
2016 5.0 1.8 2.602 0.292 0.643 0.069
2017 4.2 3.3 3.324 0.922 0.001 0.967
2018 4.2 3.1 3.324 0.578 0.050 0.967
2019 4.0 4.7 3.504 5.570 1.430 1.355
2020 3.9 3.6 3.594 1.588 0.000 1.573
Sum 52.90 23.4 13.704 3.136 10.568
Arithmetic 5.29 2.34
Mean

From the table above, we can calculate the following:

n
2
SST = ∑ (Y i − Y¯) = 13.704
i=1
n 2
SSR = ∑ (Y^ i − Y¯ ) = 10.568
i=1
n 2
SSE = ∑ (Y i − Y^i) = 3.136
i=1

248
© 2014-2024 AnalystPrep.
Measures of Goodness of Fit

We use the following measures to analyze the goodness of fit of simple linear regression:

I. Coefficient of determination.

II. F-statistic for the test of fit.

III. Standard error of the regression.

Coefficient of Determination

The coefficient of determination (R 2 ) measures the proportion of the total variability of the

dependent variable explained by the independent variable. It is calculated using the formula

below:

Explained Variation Sum of Squares Regression (SSR)

R2 = =
Total Variation Sum of Squares Total (SST)
2
∑ni=1 (Y^i − Y¯)
= 2
∑ni=1 (Y i − Y¯ )

Intuitively, we can think of the above formula as:

Total Variation − Unexplained Variation

R2 =
Total Variation
Sum of Squares Total (SST) − Sum of Squared Errors (SSE)
=
Sum of Squares Total

Simplifying the above formula gives:

Sum of Squared Errors (SSE)

R2 = 1 −
Sum of Squares Total (SST)

In the above example, the coefficient of determination is:

Explained Variation
249
© 2014-2024 AnalystPrep.
 Explained Variation
R2 =
Total Variation
Sum of Squares Regression (SSR)
=
Sum of Squares Total (SST)
10.568
= = 76.61%
13.794

Features of Coefficient of Determination (R2 )

R 2 lies between 0% and 100%. A high R 2 explains variability better than a low R2 . If R 2=1%, only

1% of the total variability can be explained. On the other hand, if R 2 =90%, over 90% of the total

variability can be explained. In a nutshell, the higher the R 2 , the higher the model's explanatory

power.

For simple linear regression (R 2 ) is calculated by squaring the correlation coefficient between

the dependent and the independent variables:

2
C ov (X, Y )
2 ∑ni=1 (Y^i − Y¯ )
r2 = R 2 = ( ) =
σX σY ∑ni=1 (Y i − Y¯ )
2

Where:

(Cov (X , Y )) = Covariance between two variables, X and Y .

(σX) = Standard deviation of X.

(σY ) = Standard deviation of Y .

Example: Calculating Coefficient of Determination (R 2 )

2 2
An analyst determines that (∑6i=1 (Yi − Y¯) = 13.704) and (∑6i=1 (Yi − Y^i ) = 3.136) from the

regression analysis of inflation rates on unemployment rates. The coefficient of determination

((R2 )) is closest to:

Solution

Sum of Squares Total (SST) − Sum of Squared Errors (SSE)

250
© 2014-2024 AnalystPrep.
 Sum of Squares Total (SST) − Sum of Squared Errors (SSE)
R2 =
Sum of Squares Total (SST)
2 2
(∑ni=1 (Yi − Y¯) − ∑ni=1 (Yi − Y^) )
13.704 − 3.136
= =
∑ni=1 (Y i − Y¯ )
2 13.704

= 0.7712 = 77.12%

F-statistic in Simple Regression Model

Note that the coefficient of determination discussed above is just a descriptive value. To check

the statistical significance of a regression model, we use the F-test, which requires us to

calculate the F-statistic.

In simple linear regression, the F-test confirms whether the slope (denoted by (b 1 )) in a

regression model is equal to zero. In a typical simple linear regression hypothesis, the null

hypothesis is formulated as: (H 0 : b1 = 0) against the alternative hypothesis (H 1 : b1 ≠ 0). The null

hypothesis is rejected if the confidence interval at the desired significance level excludes zero.

The Sum of Squares Regression (SSR) and Sum of Squares Error (SSE) are employed to calculate

the F-statistic. In the calculation, the Sum of Squares Regression (SSR) and Sum of Squares

Error (SSE) are adjusted for the degrees of freedom.

The Sum of Squares Regression(SSR) is divided by the number of independent variables (k) to

get the Mean Square Regression (MSR). That is:

2
∑ni=1 ( Ŷi − Y¯ )
SSR
MSR = =
k k

Since we only have (k = 1), in a simple linear regression model, the above formula changes to:

2
∑ni=1 ( Ŷi − Y¯) n 2
SSR
MSR = = = ∑ (Y^i − Y¯)
1 1 i =1

Therefore, in the Simple Linear Regression Model, MSR = SSR.

Also, the Sum of Squares Error (SSE) is divided by degrees of freedom given by (n − k − 1) (this

251
© 2014-2024 AnalystPrep.
translates to (n − 2) for simple linear regression) to arrive at Mean Square Error (MSE). That is,

2
Sum of Squares Error (SSE)∑i=1 (Yi − Y^)
n

MS E = =
n −k− 1 n− k− 1

For a simple linear regression model,

2
Sum of Squares Error(SSE)∑i=1 (Y i − Y^)
n

MSE = =
n −2 n −2

Finally, to calculate the F-statistic for the linear regression, we find the ratio of MSR to MSE.

That is,

2
∑n ¯
i=1 ( Ŷi − Y )
SS R
MSR k k
F − statistic = = =
MSE SS E
∑n=1 (Y i−Y^)
2

n−k−1 i
n−k−1

For simple linear regression, this translates to:

2
SS R ∑ni=1 ( Ŷi − Y¯ )
MSR k
F − statistic = = =
MSE SS E
∑n ^
2
i=1 (Y i− Y )
n−k−1
n −2

The F-statistic in simple linear regression is F-distributed with (1) and (n − 2) degrees of

freedom. That is,

MS R
∼ F1, n−2
MS E

Note that the F-test regression analysis is a one-side test, with the rejection region on the right

side. This is because the objective is to test whether the variation in Y explained (the numerator)

is larger than the variation in Y unexplained (the denominator).

Interpretation of F-test Statistic

252
© 2014-2024 AnalystPrep.
A large F-statistic value proves that the regression model effectively explains the variation in the

dependent variable and vice versa. On the contrary, an F-statistic of 0 indicates that the

independent variable does not explain the variation in the dependent variable.

We reject the null hypothesis if the calculated value of the F-statistic is greater than the critical

F-value.

It is worth mentioning that F-statistics are not commonly used in regressions with one

independent variable. This is because the F-statistic is equal to the square of the t-statistic for

the slope coefficient, which implies the same thing as the t-test.

Standard Error of Estimate

Standard Error of Estimate, Se or SEE, is alternatively referred to as the root mean square error

or standard error of the regression. It measures the distance between the observed dependent

variables and the dependent variables the regression model predicts. It is calculated as follows:

2
∑ ni=1 (Y i − Y^i )
Standard Error of Estimate (Se ) = √MSE = ⎷
n −2

The standard error of estimate, coefficient of determination, and F-statistic are the measures

that can be used to gauge the goodness of fit of a regression model. In other words, these

measures tell the extent to which a regression model syncs with data.

The smaller the Standard Error of Estimate is, the better the fit of the regression line. However,

the Standard Error of Estimate does not tell us how well the independent variable explains the

variation in the dependent variable.

Hypothesis Tests of Regression Coefficients

Hypothesis Test on the Slope Coefficient

Note that the F-statistic discussed above is used to test whether the slope coefficient is

253
© 2014-2024 AnalystPrep.
significantly different from 0. However, we may also wish to test whether the population slope

differs from a specific value or is positive. To accomplish this, we use the t-distributed test.

The process of performing the t-distributed test is as follows:

1. State the hypothesis: For instance, typical hypothesis statements include:

H 0 : b1 = 0 versus Ha : b1 ≠ 0

H 0 : b1 ≤ 0 versus Ha : b1 > 0

2. Identify the appropriate test statistic: The test statistic for the t-distributed test on

slope coefficient is given by:

^
b 1 − B1
t=
s^b1

Where: B1 = Hypothesized slope coefficient.b̂1 = Point estimate for b1 s^

b1
= Standard

error of the slope [Link] test statistic is t-distributed with n − k − 1 degrees of

freedom. Since we are dealing with simple linear regression, we will deal with n − 2

degrees of freedom. The standard error of the slope coefficient (s^

b1
) is calculated as the

ratio of the standard error of estimate (se) and the square root of the variation of the

independent variable:

se
s^
b1 =
2
√∑ni (Xi − X̄ )
=1

Where:

se = √MSE

3. Specify the level of significance: Note the level of significance level, usually denoted

by alpha, α . A typical significance level might be α = 5%

4. State the decision rule: Using the significance level, find the critical values. You can

use the t-table or spreadsheets such as Excel, statistical software such as R, or

programming languages such as Python. In an exam situation, such critical values will be

254
© 2014-2024 AnalystPrep.
 provided. Compare the t-statistic value to the critical t-value (tc) . Reject the null

hypothesis if the absolute t-statistic value is greater than the upper critical t-value or less

than the lower critical value, i.e., t > +tcritical or t < −tcritical

5. Calculate the test statistic: Using the formula above, calculate the test statistic.

Intuitively, you might need to calculate the standard error of the slope coefficient (s^
b1
)

first.

6. Make a decision: Make a decision whether to reject or fail to reject the null hypothesis.

Example: Hypothesis Test Concerning Slope Coefficient

Recall the example where we regressed inflation rates against unemployment rates from 2011 to

2020.

Year Unemployment Inflation Predicted Variation Variation Variation

Rate %(Xi ) Rate % Unemployment to be Unexplained Explained
(Y i) ^
rate (Y i ) Explained.
2 2 2
(Y i − Y¯) (Yi − Y^i ) (Y^i − Y¯ )
2011 6.1 1.7 1.610 0.410 0.008 0.533
2012 7.4 1.2 0.437 1.300 0.582 3.621
⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮
2019 4.0 4.7 3.504 5.570 1.430 1.355
2020 3.9 3.6 3.594 1.588 0.000 1.573
Sum 52.90 23.4 13.704 3.136 10.568
Arithmetic 5.29 2.34
Mean

The estimated regression model is

Y^ = 7.112 − 0.9020Xi + εi

Assume that we need to test whether the slope coefficient of the unemployment rates is positive

at a 5% significance level.

The hypotheses are as follows:

H 0 : b1 < 0 versus Ha : b1 ≥ 0

255
© 2014-2024 AnalystPrep.
Next, we need to calculate the test statistic given by:

^
b 1 −B1
t= s^
b1

Where:

se
s^
b1 =
2
√∑ni (Xi − X̄ )
=1

Recall that,

2
∑ni=1 (Y i − Y^)
SSE 3.136
se = √ MSE = √ =⎷ =√ = 0.6261
n −k− 1 n−2 8

So that,

se 0.6261
s^
b1 = = = 0.1737
√∑ni 1 (Xi − X̄ )
2 √12.989
=

Therefore,

^
b 1 − B1 −0.9020 − 0
t= = = −5.193
s^b1
0.1737

Next, we need to find critical t-values. Note that this is a one-sided test. As such, we need to find

t8 , 0.05. We will use the t-table:

256
© 2014-2024 AnalystPrep.
From the table, t8 , 0.05 = 1.860. We fail to reject the null hypothesis since the calculated test

statistic is less than the critical t-value (?5.193 < 1.860). There is sufficient evidence to indicate

that the slope coefficient is not positive.

Relationship between the Hypothesis Test of Correlation and Slope

Coefficient

In simple linear regression, a distinct characteristic exists: the t-test statistic checks if the slope

coefficient equals zero. This t-test statistic is the same as the test-statistic used to determine if

the pairwise correlation is zero.

This feature is true for two-sided tests (H 0 : ρ = 0 versus H a : ρ ≠ 0 and

H 0 : b1 = 0 versus Ha : ρ ≠ 0) and one-sided test (H 0 : ρ ≤ 0 versus Ha : ρ > 0 and

257
© 2014-2024 AnalystPrep.
H 0 : b1 ≤ 0 versus Ha : ρ > 0 or H 0 : ρ > 0 versus H a : ρ ≤ 0 and H 0 : b 1 > 0 versus H a : ρ ≤ 0).

Note that the test -statistic to test whether the correlation is equal to zero is given by:

r√ n − 2
t=
√1 − r2

The above test statistic is t-distributed with (n − 2) degrees of freedom.

Consider our previous example, where we regressed inflation rates against unemployment rates

from 2011 to 2020. Assume we want to test whether the pairwise correlation between the

unemployment and inflation rates equals zero.

In the example, the correlation between unemployment and inflation rates is -0.8782. As such,

the test- statistic to test whether the correlation is equal to zero is

−0.8782√10 − 2
t= ≈ −5.19
2
√1 − (−0.8782)

Note this is equal to the test statistic t-test statistic used to perform the hypothesis test whether

the slope coefficient is zero:

^
b 1 − B1 −0.9020 − 0
t= = = −5.193
s^b1
0.1737

Hypothesis Test of the Intercept Coefficient

Similar to the slope coefficient, we may also want to test whether the population intercept equals

a certain value. The process is similar to that of the slope coefficient. However, the test statistic

for t-distributed test on slope coefficient is given by:

^
b 0 − B0
t=
s^b0

Where:

258
© 2014-2024 AnalystPrep.
B1 = Hypothesized intercept coefficient.

b̂1 = Point estimate for b1 .

s^
b0
= Standard error of the intercept.

The formula for the standard error of the intercept s ^

b0
is given by:

2
1 X̄
s^
b0 =
+
⎷n ∑ ni=1 (Xi − X̄ )
2

Recall the example where inflation rates were regressed against unemployment rates from 2011

to 2020.

Year Unemployment Inflation Predicted Variation Variation Variation

Rate %(Xi ) Rate % Unemployment to be Unexplained Explained
(Y i) rate (Y^i ) Explained.
2 2 2
(Y i − Y¯) (Yi − Y^i ) (Y^i − Y¯ )
2011 6.1 1.7 1.610 0.410 0.008 0.533
2012 7.4 1.2 0.437 1.300 0.582 3.621
⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮
2019 4.0 4.7 3.504 5.570 1.430 1.355
2020 3.9 3.6 3.594 1.588 0.000 1.573
Sum 52.90 23.4 13.704 3.136 10.568
Arithmetic 5.29 2.34
Mean

The estimated regression model is

Y^ = 7.112 − 0.9020Xi + εi

Assume that we need to test whether the intercept is greater than 1 at a 5% significance level.

The hypotheses are as follows:

H 0 : b0 ≤ 1 versus Ha : b0 > 1

Next, we need to calculate the test statistic given by:

^
b −B
259
© 2014-2024 AnalystPrep.
 ^
b 0 − B0
t=
s^
b0

Where:

2
1 X̄ 1 5.292
s^
b0 =
+ = √ + = 1.501
⎷n ∑ni=1 (Xi − X̄ )
2 10 12.989

Therefore,

7.112 − 1
t= = 4.0719
1.501

Note that this is a one-sided test. From the table, t8 , 0.05 = 1.860. Since the calculated test

statistic is less than the critical t-value (4.0179 > 1.860), we reject the null hypothesis. There is

sufficient evidence to indicate that the intercept is greater than 1.

Hypothesis Tests Concerning Slope Coefficient When Independent

Variable is an Indicator Variable

Dummy variables, also known as indicator variables or binary variables, are used in regression

analysis to represent categorical data with two or more categories. They are particularly useful

for including qualitative information in a model that requires numerical input variables.

Example: Regression Analysis With Indicator Variables

Assume we aim to investigate if a stock's inclusion in an Environmental, Social, and Governance

(ESG) focused fund affects its monthly stock returns. In this case, we'll analyze the monthly

returns of a stock over a 48-month period.

We can use a simple linear regression model to explore this. In the model, we regress monthly

returns, denoted as R, on an indicator variable, ESG. This indicator takes the value of 0 if the

stock isn't part of an ESG-focused fund and 1 if it is.

R = b 0 + b 1 ESG + εi

260
© 2014-2024 AnalystPrep.
Note that we estimate the simple linear regression in a way similar to if the independent variable

was continuous.

The intercept β0 is the predicted value when the indicator variable is 0. On the other hand, the

slope when the indicator variable is 1 is the difference in the means if we grouped the

observations by the indicator variable.

Assume that the following table is the results of the above regression analysis:

Estimated Standard Error Calculated Test

Coefficients of Coefficients Statistic
Intercept 0.5468 0.0456 9.5623
ESG 1.1052 0.1356 9.9532

Additionally, we have the following information regarding the means and variances of the

variables.

Monthly returns Monthly Returns Difference in

of ESG Focused of Non-ESG Means
Stocks Stocks
Mean 1.6520 0.5468 1.1052
Variance 1.1052 0.1356
Observations 10 38

From the above tables, we can see that:

The intercept (0.5468) equals the mean of the returns for the non-ESG stocks.

The slope coefficient (1.1052) is the difference in means of returns between ESG-

focused stocks and non-ESG stocks.

Now, assume we want to test whether the slope coefficient equals 0 at a 5% significance level.

Therefore, the hypothesis is H 0 : β1 = 0 vs. Ha : β1 ≠ 0. Note that the degrees of freedom in

48 − 2 = 46. As such, the critical t-values (usually given in the table above) is t46, 0.025 = ±2.013.

From the first table above, the calculated test statistic for the slope is greater than the critical t-

value (9.9532 > 2.013). As a result, we reject the null hypothesis that the slope coefficient is

equal to zero.

261
© 2014-2024 AnalystPrep.
p-Values and Level of Significance

The p-value is the smallest level of significance level at which the null hypothesis is rejected.

Therefore, the smaller the p-value, the smaller the probability of rejecting the true null

hypothesis (type I error) and, hence, the greater the validity of the regression model.

Software packages commonly offer p-values for regression coefficients. These p-values help test

a null hypothesis that the true parameter equals 0 versus the alternative that it's not equal to

zero.

We reject the null hypothesis if the p-value corresponding to the calculated test statistic is less

than the significance level.

Example: Hypothesis Testing of Slope Coefficients

An analyst generates the following output from the regression analysis of inflation on

unemployment:

Regression Statistics
R Square 0.7684
Standard Error 0.0063
Observations 10

Coefficients Standard Error t-Stat

Intercept 0.0710 0.0094 7.5160
Forecast (Slope) −0.9041 0.1755 −5.1516

At the 5% significant level, test the null hypothesis that the slope coefficient is significantly

different from one, that is,

H0 : b 1 = 1 vs. H a : b1 ≠ 1

Solution

^
b 1 −b1
The calculated t-statistic, t = is equal to:
^
S b1

−0.9041 − 1
t= = −10.85
0.1755

262
© 2014-2024 AnalystPrep.
The critical two-tail t-values from the table with n − 2 = 8 degrees of freedom are:

tc = ±2.306

Notice that | t| > tc i.e., (10.85 > 2.306)

Therefore, we reject the null hypothesis and conclude that the estimated slope coefficient is

statistically different from one.

Note that we used the confidence interval approach and arrived at the same conclusion.

263
© 2014-2024 AnalystPrep.
Question 1

Samantha Lee, an investment analyst, is studying monthly stock returns. She focuses

on companies listed in a Renewable Energy Index across various economic

conditions. In her analysis, she performed a simple regression. This regression

explains how stock returns vary concerning the indicator variable RENEW. RENEW

equals 1 when there's a positive policy change towards renewable energy during that

month, and 0 if not. The total variation in the dependent variable amounted to

220.34. Of this, 94.75 is the part explained by the model. Samantha's dataset includes

36 monthly observations.

Calculate the coefficient of determination, F-statistic, and standard deviation of

monthly stock returns of companies listed in a Renewable Energy Index.

A. R 2=43.00%;F=26.07;Standard deviation=2.51.

B. R 2=53.00%;F=26.41;Standard deviation=2.55.

C. R 2=33.00%;F=36.07;Standard deviation=3.55.

Solution

The correct answer is A.

Coefficient of determination:

Explained variation 94.75

R2 = = ≈ 43%
Total variation 220.34

F-statistic:

Explained variation SS R 94.75

k k 1
F= = = = 26.07
Unexplained variation SS E 220.34−94.75
n−2 n−2 34

Standard deviation:

Note that,

264
© 2014-2024 AnalystPrep.
 n
2
Total Variation = ∑ (Yi − Y¯) = 220.34
i=1

And the standard deviation is given by:

2
∑ni=1 (Y i − Y¯ )
Standard deviation = ⎷
n−1

As such,

Total variation 220.34

Standard deviation = √ =√ = 2.509
n− 1 n −1

Question 2

Neeth Shinu, CFA, is forecasting the price elasticity of supply for a specific product.

Shinu uses the quantity of the product supplied for the past 5months as the

dependent variable and the price per unit of the product as the independent variable.

The regression results are shown below.

Regression Statistics
R Square 0.9941
Standard Error 3.6515
Observations 5
Coefficients Standard Error t Stat P-value
Intercept −159 10.520 (15.114) 0.001
Slope 0.26 0.012 22.517 0.000

Which of the following most likely reports the correct value of the t-statistic for the

slope and most accurately evaluates its statistical significance with 95% confidence?

A. t = 21.67; the slope is significantly different from zero.

B. t = 3.18; the slope is significantly different from zero.

C. t = 22.57; the slope is not significantly different from zero.

Solution

265
© 2014-2024 AnalystPrep.
The correct answer is A.

The t-statistic is calculated using the formula:

^
b 1 − b1
t=
S^b 1

Where:

b1 = True slope coefficient.

^
b 1 = Point estimator for B 1 .

S^b 1 = Standard error of the regression coefficient.

0.26 − 0
t= = 21.67
0.012

The critical two-tail t-values from the t-table with n − 2 = 3 degrees of freedom are:

tc = ±3.18

266
© 2014-2024 AnalystPrep.
Notice that | t| > tc (i.e., 21.67 > 3.18).

Therefore, the null hypothesis can be rejected. Further, we can conclude that the

estimated slope coefficient is statistically different from zero.

267
© 2014-2024 AnalystPrep.
LOS 10d: describe the use of analysis of variance (ANOVA) in regression
analysis, interpret ANOVA results, and calculate and interpret the
standard error of estimate in a simple linear regression

The sum of squares of a regression model is usually represented in the Analysis of Variance

(ANOVA) table. The ANOVA table contains the sum of squares (SST, SSE, and SSR), the degrees

of freedom, the mean squares (MSR and MSE), and F-statistics.

The typical format of ANOVA is as shown below:

Source Sum of Squares Degrees Mean F-statistic

of square
Freedom
2
SS R MS R
Regression SSR = ∑ ni=1 ( Ŷ i − Y¯) 1 MSR = F =
1 MS E
(Explained)
2
SS E
Residual SS E = ∑ni=1 (Yi − Ŷi ) n− 2 MSE =
n −2
(explained)
2
Total SST = ∑ni=1 (Yi − Y¯) n− 1

Standard Error of EstimateStandard Error of Estimate, Se or SEE, is referred to as the root

mean square error or standard error of the regression. It measures the distance between the

observed and dependent variables predicted by the regression model. The Standard Error of

Estimate is easily calculated from the ANOVA table using the following formula:

2
∑ ni=1 (Y i − Y^)
Standard Error of Estimate (Se ) = √MSE = ⎷
n− 2

The standard error of estimate, coefficient of determination, and F-statistic are the measures

that can be used to gauge the goodness of fit of a regression model. In other words, these

measures are used to tell the extent to which a regression model syncs with data.

The smaller the Standard Error of Estimate is, the better the fit of the regression line. However,

the Standard Error of Estimate does not tell us how well the independent variable explains the

variation in the dependent variable.

268
© 2014-2024 AnalystPrep.
Example: Calculating and Interpreting F-Statistic

The completed ANOVA table for the regression model of the inflation rate against the

unemployment rate over 10 years is given below:

Source Sum of Degrees of Mean Sum F-Statistic

Squares Freedom of Squares
Regression 10.568 1 10.568 ?
Error 3.136 8 0.392
Total 13.704 9

a. Use the above ANOVA table to calculate the F-statistic.

b. Test the hypothesis that the slope coefficient equals a 5% significance level.

Solution

Mean Regression Sum of Squares (MSR)

10.568
a. = = 0.392 = 26.960
Mean Squared Error(MSE)

b. We are testing the null hypothesis H 0 : b1 = 0 against the alternative hypothesis

H 1 : b1 ≠ 0. The critical F-value for k = 1 and n − 2 = 8 degrees of freedom at a 5%

significance level is roughly 5.32. Note that this is a one-tail test, so we use the 5% F-

table.

269
© 2014-2024 AnalystPrep.
Remember that the null hypothesis is rejected if the calculated value of the F-statistic is

greater than the critical value of F. Since 26.960 > 5.32, we reject the null hypothesis

and conclude that the slope coefficient is significantly different from zero. Notice that we

also rejected the null hypothesis in the previous examples. We did so because the 95%

confidence interval did not include [Link] F-test duplicates the t-test in regard to the

slope coefficient significance for a linear regression model with one independent

variable. In this case, t2 = 2.3062 ≈ 5.32. Since the F-statistic is the square of the t-

statistic for the slope coefficient, its inferences are the same as the t-test. However, this

is not the case for multiple regressions.

270
© 2014-2024 AnalystPrep.
Question

Consider the following analysis of variance (ANOVA) table:

Source Degrees of Sum of Mean Sum

Freedom Squares of Squares
Regression 1 1, 701, 563 1 , 701, 563
Error 3 106, 800 13 , 350
(Unexplained)
Total 4 1, 808, 363

The value of R 2 and the F-statistic for the test of fit of the regression model are

closest to:

A. 6% and 16.

B. 94% and 127.

C. 99% and 127.

Solution

The correct answer is B.

Sum of Squares Regression (SSR)

1, 701, 563
R2 = = = 0.94 = 94%
Sum of Squares Total (SST) 1, 808, 363

Mean Regression Sum of Squares (MSR)

F=
Mean Squared Error (MSE)
1, 701, 563
= = 127.46 ≈ 127
13, 350

271
© 2014-2024 AnalystPrep.
LOS 10e: calculate and interpret the predicted value for the dependent
variable, and a prediction interval for it, given an estimated linear
regression model and a value for the independent variable

We calculate the predicted value of the dependent variable, Y , by inserting the estimated value

of the independent variable, X , into the regression equation. The predicted value of the

dependent variable, Y , is determined using the following formula:

Y^ = ^
b0 +^
b 1X

Where:

Y^ = Predicted value of the dependent variable.

X = Estimated value of the independent variable.

Example: Calculating the Predicted Value of a Dependent Variable

Refer to the example of regressed inflation rates against unemployment rates from 2011 to 2020.

Year Unemployment Inflation Predicted Variation Variation Variation

Rate %(Xi ) Rate % Unemployment to be Unexplained Explained
(Y i) ^
rate (Y i ) Explained.
2 2 2
(Y i − Y¯) (Yi − Y^i ) (Y^i − Y¯ )
2011 6.1 1.7 1.610 0.410 0.008 0.533
2012 7.4 1.2 0.437 1.300 0.582 3.621
⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮
2019 4.0 4.7 3.504 5.570 1.430 1.355
2020 3.9 3.6 3.594 1.588 0.000 1.573
Sum 52.90 23.4 13.704 3.136 10.568
Arithmetic 5.29 2.34
Mean

The estimated regression model is illustrated below.

Y^ = 7.112 − 0.9020Xi + εi

Calculate the predicted inflation rate value if the forecasted value of the unemployment rate is

272
© 2014-2024 AnalystPrep.
4.5%.

Solution

The predicted value of the inflation rate is determined as follows:

Y^ = 7.112 − 0.9020 × 4.5 = 3.053%

Confidence Interval for Predicted Values

The confidence interval calculation for the predicted value of a dependent variable is the same as

that of the confidence interval for regression coefficients. The confidence interval for a predicted

value of the dependent variable is given by:

Prediction Interval = Y^ ± tcsf

Where:

tc = Two-tailed critical t-value at the given significance level with n − 2 df.

Y^ = Predicted value of a dependent variable.

s2f = The estimated variance of the prediction error.

2 2
⎡ 1 (Xf − X̄) ⎤ ⎡ 1 (Xf − X̄) ⎤
s2f = s2e 1+ + = s2e 1 + +
⎣ n (n − 1) sx ⎦
2 ⎣ n 2
∑ni = 1 (X i − X̄) ⎦

Where:

s2e = The squared standard error of the estimate.

n = Number of observations.

s2X = Variance of the independent variable.

Xf = Value of the independent variable.

273
© 2014-2024 AnalystPrep.
We can, therefore, calculate the standard error of forecast as shown below:

2
1 (Xf − X̄ )
sf = se 1 + +
⎷ n 2
∑ni=1 (Xi − X̄ )

From the formula above, we can observe that:

A better fit of the regression analysis leads to a smaller standard error of the estimate

(se ), subsequently resulting in a lower standard error of the forecast.

When the sample size (n) in the regression calculation increases, it directly

corresponds to a reduction in the standard error of the forecast.

If the forecasted independent variable (X f ) approaches the mean of the independent

variable (X̄ ) utilized in the regression analysis, it decreases the standard error of the

forecast.

Example: Calculating the Confidence Interval of the Predicted Value

Refer to the example of regressed inflation rates against unemployment rates from 2011 to 2020.

Year Unemployment Inflation Predicted Variation Variation Variation

Rate %(Xi ) Rate % Unemployment to be Unexplained Explained
(Y i) ^
rate (Y i ) Explained.
2 2 2
(Y i − Y¯) (Yi − Y^i ) (Y^i − Y¯ )
2011 6.1 1.7 1.610 0.410 0.008 0.533
2012 7.4 1.2 0.437 1.300 0.582 3.621
⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮
2019 4.0 4.7 3.504 5.570 1.430 1.355
2020 3.9 3.6 3.594 1.588 0.000 1.573
Sum 52.90 23.4 13.704 3.136 10.568
Arithmetic 5.29 2.34
Mean

Consider the results of the regression analysis of inflation rates on unemployment rates:

274
© 2014-2024 AnalystPrep.
 Regression Statistics
R Square 0.7711
Standard Error 0.6261
Observations 10

ANOVA
df Sum of Mean F
Squares Square
Regression 1 10.568 10.568 26.9565
Residual 8 3.136 0.392
Total 9 13.704

Coefficients Standard t Stat p-value

Error
Intercept 7.112 0.940 7.565 0.000
Unemployment −0.902 0.174 −5.192 0.001
rate (%)

Given that the forecasted unemployment rate is 4.5%, calculate the 95% confidence interval for

the predicted inflation rate value.

Solution

Prediction Interval = Y^ ± tcs f

The estimated variance of the prediction error is:

2
⎡ 1 (X f − X̄ ) ⎤
s 2f = s 2e 1 + +
⎣ n (n − 1) s2X ⎦
2
⎡ 1 (X f − X̄ ) ⎤
= s 2e 1+ +
⎣ n 2
∑ni=1 (Xi − X̄ ) ⎦
2
1 (4.5 − 5.29)
= 0.62612 [1 + + ] = 0.450
10 12.989

As such, the standard error of forecast is:

sf = √0.450 = 0.6708

The predicted value of the inflation rate given an unemployment rate of 4.5% is:

275
© 2014-2024 AnalystPrep.
 Y^ = 7.112 − 0.9020 × 4.5 = 3.05%

The two-tailed critical t-value with 8 (n − 2) degrees of freedom at the 5% significance level is

2.306.

The prediction interval at the 95% confidence level is:

Prediction Interval (PI) = Y^ ± tcsf

276
© 2014-2024 AnalystPrep.
 PI = 3.05 ± 2.306 × 0.6708 = 1.50% to 4.60%

Interpretation

Given an unemployment rate of 4.5%, we are 95% confident that the inflation rate will lie

between 1.50% and 4.60%.

277
© 2014-2024 AnalystPrep.
Question 1

The regression equation of the quantity of goods against the price is given by:

Y = −159 + 0.26X

Where:

Y = Quantity supplied.

X = Price per unit of the product.

The predicted value of the quantity supplied when the price equals 1,200 is closest

to:

A. 153.

B. 155.

C. 471.

The correct answer is A.

Y = −159 + 0.26 × 1, 200 = 153

278
© 2014-2024 AnalystPrep.
LOS 10f: Describe different functional forms of simple linear regressions

To address non-linear relationships, we employ various functional forms to potentially convert

the data for linear regression. Here are three commonly used log transformation functional

forms:

1. Log-lin model: In this log transformation, the dependent variable is logarithmic, while

the independent variable is linear. It is represented as shown below.

lnY = b0 + b1 Xi .

The slope coefficient in the log-lin model is the relative change in the dependent variable

for an absolute change in the independent variable.

When utilizing a log-lin model, caution must be exercised when making forecasts. For

example, in the predicted regression equation like Y = −3 + 5X, if X is equal to 1, the

lnY = −3 , then,

Y = e−3 = 0.0498

Moreover, the lin-lin model cannot be compared with the log-lin model without the

transformation. As such, we need to transform R2 and F-statistic.

2. Lin-log model: In this case, the dependent variable is linear, while the independent

variable is logarithmic. It is represented as follows:

Y i = b 0 + b 1 lnXi .

The slope coefficient in the lin-log model is responsible for the absolute change in the

dependent variable for a relative change in the independent variable.

3. Log-log model: In this log transformation, both the dependent and independent

variables are logarithmic. It is represented as lnYi = b0 + b 1 lnXi . The slope coefficient in

the log-log model is the relative change in the dependent variable for a relative change in

the independent variable. In other words, if X increases by 1%, Y will change by b1 .

279
© 2014-2024 AnalystPrep.
Selecting the Correct Functional Form

To settle on the correct functional form, consider the following goodness of fit measures:

I. Coefficient of determination (R 2 ). A high value is better.

II. F-statistic. The high value of the F-statistic is better.

III. Standard error of the estimate (Se ). A low value of Se is better.

In addition to the factors cited above, the patterns in residuals can also be analyzed when

evaluating a model. In a good model, residuals are random and uncorrelated.

280
© 2014-2024 AnalystPrep.
Question 1

Which of the following statements about the log-lin model is most likely correct:

A. The dependent variable is linear, while the independent variable is

logarithmic.

B. Both the dependent and independent variables are logarithmic

C. The dependent variable is logarithmic, while the independent variable is

linear.

The correct answer is c.

In the log-lin model, the dependent variable (Y ) is logarithmic, as represented by

lnY = b 0 + b 1 Xi

While the independent variable (X ) is linear.

A is incorrect. It describes the lin-log model, where the dependent variable is linear

and the independent variable is logarithmic.

B is incorrect. It describes the log-log model, where both the dependent and

independent variables are logarithmic.

281
© 2014-2024 AnalystPrep.
 Learning Module 11: Introduction to Big Data Techniques

LOS 11b: describe Big Data, artificial intelligence, and machine learning

Big data is a term that describes large, complex datasets. These datasets are analyzed with

computers to uncover patterns and trends, particularly those related to human behavior. Big

data includes traditional sources like company reports and government data and non-

traditional sources like social media, sensors, electronic devices, and data generated as a

byproduct of a company's operations.

Characteristics of Big Data

Volume: The amount of data collected in various forms, including files, records, tables, etc.

Quantities of data reach almost incomprehensible proportions.

Velocity: The speed of data processing can be extremely high. In most cases, we deal with real-

time data.

Variety: The number of types/formats of data. The data could be structured (e.g., SQL tables or

CSV files), semi-structured (e.g., HT ML code), or unstructured (e.g., video messages).

Veracity: This is the trustworthiness and reliability of data sources. Veracity is crucial when

using big data for making predictions or drawing conclusions. Big data makes it challenging to

distinguish between data quality and quantity.

Types of Big Data

Big Data can be structured, unstructured, or semi-structured:

Structured data refers to information with a high degree of organization. Items can be

organized in tables and stored in a database where each field represents the same type of

information.

282
© 2014-2024 AnalystPrep.
Unstructured data refers to information with a low degree of organization. Items such as text

messages, tweets, emails, voice recordings, pictures, blogs, scanners, and sensors are

unorganized and cannot be presented in tabular form.

Semi-structured data may have the qualities of both structured and unstructured data.

Sources of Big Data

Financial markets: Equity, swaps, futures, options, and other derivatives.

Businesses: Financial statements, credit card purchases, and commercial

transactions.

Governments: Payroll, economic, trade, employment data, etc.

Individuals: Product reviews, credit card purchases, social media posts, etc.

Sensors: Shipping cargo information, traffic data, and satellite imagery.

The Internet of Things: data generated by ‘smart’ buildings through fittings such as

CCT V cameras, vehicles, home appliances, etc.

Professional investors, particularly quantitative ones, use alternative data sources in their

financial analysis and decision-making. These sources significantly influence how they conduct

their processes. They use alternative data to support data-driven investment models and

decisions.

The following are the top three alternative data sources:

People-generated data: This data is unstructured and is primarily accessed through

website clicks and page visits

Commercial operations data: This includes data on credit cards and corporate

exhaust. It includes information from business transactions like point-of-sale records

and banking activities. This data is typically structured.

283
© 2014-2024 AnalystPrep.
 Data produced by sensors: This data is typically unstructured and is gathered

through satellites, smartphones, cameras, RFID chips, and webcams.

Investment professionals must consider legal and ethical aspects when they use non-public

information. Web data scraping can gather personal data that might be legally protected or

disclosed without people's knowledge or consent.

Big Data Challenges

Quality: Important questions include, but are not limited to, "Does the dataset contain

selection bias, missing data, or outliers?"

Volume: Is the quantity of data gathered adequate?

Appropriateness: Is the dataset suitable for the chosen analysis method?

Experts have created artificial intelligence (AI) and machine learning methods to handle large

and intricate alternative datasets. These technologies help in understanding and evaluating this

vast and complex data.

Artificial Intelligence (AI) and Machine Learning (ML)

Artificial Intelligence

In broad terms, artificial intelligence refers to machines that can perform tasks in “intelligent”

ways. It has much to do with developing computer systems that exhibit cognitive and decision-

making abilities comparable to or superior to humans. It is the broader concept of machines

being able to carry out tasks in a way that we would consider “smart.”

Early AI took the shape of expert systems, using "if-then" computer programming to mimic

human knowledge and analysis. Neural networks, another early form, mimicked human brain

functions in learning and processing information.

284
© 2014-2024 AnalystPrep.
Machine Learning

Machine learning is a current application of AI that revolves around the idea that we should

really just give machines access to data and let them learn by themselves without making any

assumptions about the underlying probability distribution.

The idea is that when exposed to more data, machines can make changes independently and

come up with solutions to problems without reliance on human expertise – find and apply the

pattern.

In the context of investment, machine learning requires big data for training. The growth of big

data has enabled AI algorithms to improve modeling and predictive accuracy.

In machine learning (ML), a computer algorithm receives inputs, which can be datasets or

variables, as well as outputs, representing the target data. The algorithm then learns how to

effectively model inputs into outputs or describe a data structure. It learns by identifying data

relationships and using this knowledge to enhance its learning process.

The ML divides the dataset into three unique types: a training dataset, a validation dataset,

and a test dataset. A training dataset allows the algorithm to identify the link between inputs

and outputs based on the historical pattern in the data. These relationships are then validated,

and the model is adjusted using the validation dataset.

As the name suggests, the test dataset tests the model's strength in predicting well on the new

data. Note that machine learning still needs human intervention to understand the underlying

data and choose suitable techniques for data analysis. In other words, before data is utilized, it

must be cleaned and free from bias and spurious data.

Causes of Errors in Machine Learning

Overfitting the Data

The model overfits the data when it discovers “false” associations or “unsubstantiated” patterns

that cause prediction errors and wrong forecasts. In other words, overfitting happens when the

ML model is overtrained on the data and considers the noise in the data as true parameters.

285
© 2014-2024 AnalystPrep.
Underfitting the Data

Underfitting of data occurs when the model considers the true parameters as noise and is unable

to identify the relationship within the training data. In other words, the model is too simple to

recognize patterns in the data.

Black Box Problem

Machine learning models don't use explicit rules like traditional software. They learn from lots of

data during training. This makes ML models, such as black boxes, sometimes give results that

are hard to understand or describe.

Types of Machine Learning

Supervised Learning

Under supervised learning, computers learn to model data based on labeled training data

containing inputs and the desired outputs. After “learning” how best to model the relationships

for the labeled data, the algorithms are employed to predict the results for the new datasets.

Unsupervised Learning

In unsupervised learning, computers get input data without labels and have to describe it, often

by grouping data points. They learn from unlabeled data and react based on commonalities. For

example, grouping companies based on their financial, not geographical or industrial,

characteristics is unsupervised learning.

Deep Learning

Deep learning involves computers using neural networks to process data in multiple stages,

identifying complex patterns. It employs both supervised and unsupervised machine learning

methods.

286
© 2014-2024 AnalystPrep.
Question

Machine learning refers to one of the following:

A. Autonomous acquisition of knowledge through the use of computer programs.

B. Ability of machines to execute coded instructions.

C. Selective acquisition of knowledge through the use of computer programs.

Solution

The correct answer is A.

Machine learning means computers independently acquire knowledge through

programs, enabling them to solve problems without human input. It's about

computers learning and making decisions on their own.

287
© 2014-2024 AnalystPrep.
LOS 11c: Describe applications of Big Data and Data Science to
investment management

Data science is an interdisciplinary field that uses developments in computer science, statistics,

and other fields to extract information from Big Data or data in general.

Data Processing Methods

Data analysts and scientists in big data analysis use different data management approaches.

They consist of capture, curation, storage, search, and transfer.

Capture: Describes the method by which data is gathered and put into a form that the

analytical process may use.

Curation: Data curation ensures the quality and accuracy of the data by undertaking a

data cleaning activity. This procedure finds data inaccuracies, and any missing data is

compensated for.

Storage: Process of recording, archiving, and accessing data, as well as the

fundamental structure of the underlying database:

Search: Involves querying data to locate specific information. With big data,

sophisticated techniques are necessary to efficiently retrieve the requested data

content.

Transfer: Describes the process of transferring data from the underlying data source

or storage place to the underlying analytical instrument.

Data Visualization

Visualization encompasses data formatting, display, and summarization through graphical

representations. Tables, charts, and trends are commonly used for traditional structured data,

while non-traditional unstructured data demands innovative techniques like interactive three-

288
© 2014-2024 AnalystPrep.
dimensional (3D) graphics, tag clouds, and mind maps.

Fintech is applied in investment management, including text analytics, natural language

processing, risk assessment, and algorithmic trading.

Text Analytics and Natural Language

Text analytics employs computer programs to analyze and extract insights, primarily from

unstructured text- or voice-based datasets like company filings, written reports, quarterly

earnings calls, and social media content. Text analytics can be utilized in predictive analysis to

identify potential indicators of future performance, such as consumer sentiment.

Natural language processing (NLP) is an area of study that involves creating computer programs

to decipher and analyze human language. Essentially, NLP combines computer science, AI, and

linguistics.

Translation, speech recognition, text mining, sentiment analysis, and topic analysis are examples

of automated tasks that use NLP. Annual reports, call transcripts, news articles, social media

posts, and other text- and audio-based data may all be analyzed using natural language

processing (NLP), allowing NLP to discover trends more quickly and accurately than is humanly

possible.

Using natural language processing data, earnings projections for a company's near-term

prospects can be created. X (formerly Twitter) sentiments have also been used to gauge an initial

public offering (IPO) success.

Python, R, and Excel VBA are frequently used programming languages, whereas SQL, SQLite,

and NoSQL are prominent database systems.

289
© 2014-2024 AnalystPrep.
Question

Which of the five data processing methods refers to the process of ensuring data

quality and accuracy through a data cleaning exercise?

A. Data search.

B. Data storage.

C. Data curation.

The correct answer is C.

Data curation refers to the process of ensuring data quality and accuracy through a

data cleaning exercise. It involves uncovering data errors and adjusting for missing

data.

A is incorrect. Data search refers to how to query data. Big data requires advanced

techniques to locate requested data content.

B is incorrect. Data storage refers to how the data will be recorded, archived, and

accessed and the underlying database design.

290
© 2014-2024 AnalystPrep.
LOS 11a: describe aspects of “fintech” that are directly relevant for the
gathering and analyzing of financial data

Fintech refers to technological innovation in designing and delivering financial services and

products. At its core, fintech has helped companies, business owners, and investment managers

better manage their operations through specialized software and algorithms.

Note that the term fintech is commonly used to refer to companies that develop new

technologies and their applications and also the business sector that encompasses such

companies.

Fintech in Gathering and Analyzing Financial Data

Initially, financial innovation was limited to simple tasks such as data processing and automation

of routine tasks. Today, fintech encompasses more advanced systems that can analyze

information and make decisions based on machine-learning logic. Machines have been developed

to “learn” how to perform tasks over time. Using such systems has brought about high levels of

efficiency that surpass human capabilities. Fintech covers a broader range of services and

applications. As such, services and applications of fintech relevant to the investment industry

include:

1. Analysis of Large Datasets: Apart from traditional data such as corporate financial

statements and economic indicators, fintech development has helped integrate

alternative data, such as social media, into investment decision-making.

2. Analytical Tools: Artificial Intelligence (AI) can identify complex, non-linear

relationships compared to traditional quantitative methods by enabling different data

analysis techniques. Diverse approaches to data analysis are now possible because of

advancements in AI-based methodologies. As an illustration, analysts use AI to sift

through the vast volumes of data from corporate filings and annual reports to produce

insights.

291
© 2014-2024 AnalystPrep.
Question

A characteristic of fintech is that it is:

A. at its most advanced state, using systems that follow specified rules and

instructions.

B. limited to simple tasks such as automating routine processes and data processing.

C. primarily driven by the increased availability of data and technological

advancement.

The correct answer is C.

The availability of vast amounts of data and technological advancements have been

the primary drivers of fintech's expansion. The rapid growth in data, including

diverse types, large quantities, and improved quality, has provided valuable insights

for financial institutions and fintech companies to develop data-driven solutions.

Technological advancements, such as artificial intelligence and big data analytics,

have made it possible to analyze and interpret this data effectively, creating

innovative financial products and services.

A is incorrect. While this may be true for some aspects of fintech, it doesn't directly

address the two most important reasons behind its growth - the rapid growth in data

and technological advancements. The advanced state of fintech is more a result of

leveraging data and technological innovations to develop sophisticated and efficient

financial solutions.

B is incorrect. Fintech is not limited to simple tasks; it has expanded to encompass

various complex financial activities. While it does automate routine processes and

data processing, it does so through advanced technologies that enable the handling of

massive amounts of data efficiently.

292
© 2014-2024 AnalystPrep.