NORTHEASTERN UNIVERSITY

COLLEGE OF BUSINESS ADMINISTRATION

M aster of Science in F inance

Review Module #3

STATISTICAL AND MATHEMATICAL METHODS

I. BASIC MATHEMATICAL METHODS

 LEARNING OBJECTIVES

 Be Able to Perform Mathematical Operations on Variables and Exponents

 Demonstrate the Ability to Solve and Interpret Equations

 Be Able to Identify the Slope and the Intercept of a Linear Function

 Understand Functions and the Differences Between Linear Functions and

Non-linear Functions

 Be Able to Classify a Function as Either Concave or Convex for a Particular

Set of Input Variables
 THE BASICS

Although most students have a pretty good understanding of basic mathematical

methods, its always pays to review fundamental techniques. Even Michael
Jordan practices layups!

 A Variable
 A quantity that may assume a succession of values
 A symbol standing for any one of a class of things [Webster’s]
 Fundamental building block used in the construction of algebraic
equations

 So, how do variables work?

 Multiplication
xn * xm= xn+m
i.e., x*x = x1+1 = x2
x4 * x2 = x4+2 = x6
2 * 22 = 21+2 = 23 = 8
 Division
xn / xm = xn-m
i.e., x2 / x = x2-1 = x
x6 / x2 = x6-2 = x4
2 / 22 = 21-2 = 2-1 = ½
 Exponential
(xn)m = xn*m
i.e., (x)2 = x1*2 = x2
(x4 )2 = x4*2 = x8
(2x)2 = 22 x2 = 4x2
(x2 / y3)2 = x2*2 / y3*2 = x4 / y6
 EQUATIONS

Financial and economic principles are often expressed (or modeled) as

mathematical equations. If we view mathematics as another language that we
use to express economic ideas, we can think of variables as words and equations
as sentences. To understand finance, we need to understand equations.

 Solving Equations

 When solving algebraic equations, simplify using basic rules

or techniques:
addition
subtraction
multiplication
division
consolidation
expansion
substitution
 Can add the same quantity to both sides of an equation:
y - 2 = 2x + 4
y - 2 + 2 = 2x + 4 + 2
then
y = 2x + 6
 Can subtract the same quantity from both sides:
y+2=x+2
y+2-2=x+2-2
then
y=x
 Can multiply both sides:
y/2 = x + 2
2(y/2) = 2(x + 2)
then
y = 2x + 4
 Can divide both sides:

4y = 3x + 12
4y/4 = (3x +12)/4
y = 3/4x + 12/4
then
y = 3/4x + 3
 Can consolidate variables into a more manageable form:
y - x = 2x + 3x - 7
y - x + x = 2x + 3x + x - 7
y = x(2 + 3 + 1) - 7
then
y = 6x - 7
 Can expand the equation:
y = x(1 + x) - 6
y = x + x2 - 6
then
y = x2 + x - 6
Further, y = x2 + x - 6 can be factored:
y = (x + 3)(x - 2)
example: What if ‘someone’ wanted to know, for what values does y
= 0?
solution: y = 0 when (x + 3) = 0 or when (x - 2) = 0
x+3=0
x+3-3=0-3
then
x = -3
or
x-2=0
x -2+2=0+2
then
x=2
 Can substitute equations into each other:
example #1:
y = 4x + 2
x=4
then
y = 4 (4) + 2
y = 16 + 2
y = 18
 Substitution -- continued
example #2:
y = 5x + 12
y = 7x - 2

Since both equations are equal to ‘y’, we can set the two equations
equal to each other:
7x -2 = 5x + 12
7x - 5x - 2 + 2 = 5x - 5x + 12 + 2
7x - 5x = 12 + 2
2x = 14
then
x=7
Further, what is the value of ‘y’ when x=7?
y = 5(7) + 12
y = 35 + 12
then
y = 47
or
y = 7(7) - 2
y = 49 -2
then
y = 47

 Interpreting Equations

 So far we have dealt with equations of the form:

y = mx + b

where m and b are numbers.

This formulation is known as a linear equation.

  The linear equation (the equation of a line)

m ≡ slope = Δy / Δx

and represent s the change in the value of ‘y’ for a ‘unit’ change in ‘x’.

i.e., if Δy = 3 when Δx = 1 ⇒ m = 3/1 = 3

if Δy = 4 when Δx = 2 ⇒ m = 4/2 = 2

b ≡ intercept term

and represents the value of ‘y’ when x = 0. Thus, the value of ‘y’ as
a
function of ‘x’ is shifted by the value of ‘b’.

example:
y=x
y=x+2
y = 3x + 4

Variable Values

x 0 1 2 3 4

y=x 0 1 2 3 4

y=x+2 2 3 4 5 6
y = 3x + 4 4 7 10 13 16

for y = x
the slope = 1
the intercept = 0
for y = x + 2
the slope = 1
the intercept = 2
for y = 3x + 4
the slope = 3
the intercept = 4
 Functions

We could alternatively express their relationships in the last example as:

y = f(x)
or
y is a function of x
 A function is a ‘rule’ that assigns a unique number (or value ‘y’) to
each value of ‘x’
values of ‘x’ are called the Domain (the set of permissible
inputs)
values of ‘y’ are called the Range (the outputs)

 A function is said to ‘map’ from the domain to the range

Given all this, when will I ever use this in my life (other than Math class)?
Answer: YES! Here are a few examples.

(1) The Martini Function

MARTINI:

Recipe: 1 part Vermouth

5 parts Gin

Add to Ice
Stir or Shake

or
M = V + 5G

(2) The Total Cost Function

Total Cost = Variable Cost + Fixed Costs
TC = VC*q + FC

(3) The Basic Accounting Function

Total Assets = Total Liabilities + Owners’ Equity
A=L+E

(4) The Capital Asset Pricing Model

The Return on a Stock = Beta*(Market Premium) + Risk Free Rate
 R stock = β(R market - R risk free ) + R risk free

 Non-linear Functions

Functions that are not of the form y = mx + b are known as non-linear

functions

examples: y = ax2 + bx + c (quadratic function)

y = ax3 + bx2 + cx + d (cubic function)
y = 1/x (inverse function)

numerical example:
y = 2x2 + 3x + 1
y = x3 + 2x2 + 4x
y = 1/x

Variable Values

x 0 1 2 3 4

y = 2x2 + 3x + 1 1 6 15 28 45

y = x3 + 2x2 + 4x 0 7 32 57 112
y = 1/x * 1 0.5 0.333 0.25

 Comments

1. The quadratic function and the cubic function each have an intercept
(y takes a value when x=0) but the inverse function does not [y is
undefined or is defined to be infinity (y=∞)].
2. x=0 is not in the domain of the inverse function: it is not a permissible
input.
3. Note that the change in y for a given change in x is no longer
constant as in the linear examples. Non-linear functions have
an instantaneous slope, which is a slope only for a given value of x.
The instantaneous slope is obtained by evaluating a function’s
derivative for a particular value of x.
 Concavity and Convexity

Consider the following two functions:

y = x3
y = 0.5x0.5

Variable Values

x 0 1 2 3 4

y = x3 0 1 8 27 64

Δy NA 1 7 19 37

y = 0.5x0.5 0 0.5 0.71 0.87 1

Δy NA 0.5 0.21 0.16 0.13

 Note that y = x3 is increasing at an increasing rate [e.g., when x

increases from 0 to 1, y increases from 0 to 1 (Δy = 1), but as x
increases from 1 to 2 y increases from 1 to 8 (Δy = 7)]. A function
that increases at an increasing rate is said to be convex (some math
texts call this concave up).
 Note that y = 0.5x0.5 is increasing at a decreasing rate [e.g., when x
increases from 0 to 1, y increases from 0 to 0.5 (Δy = 0.5), but as x
increases from 1 to 2 y increases from 0.5 to 0.71 (Δy = 0.21)]. A
function that increases at a decreasing rate is concave (some math
texts call this concave down).
Comment: In economics we often use concave functions to
represent utility (a function that measures the preferences of people)
example #1: Investor ‘A’ has a portfolio worth $10,000,000 and
investor ‘B’ has a portfolio worth $100,000. Both investors receive a
$100,000 gain on their portfolios in a given period. It stands to
reason that both investors prefer a gain (utility is increasing). On
the other hand, it also stands to reason that ‘B’ has a greater
preference for the gain since B’s portfolio doubles (goes from
$100,000 to $200,000) and A’s portfolio only increases by 1% (goes
from $10,000,000 to $10,100,000). Thus, we argue that investors
utility is concave in wealth -- that is, it is always increasing, but at a
decreasing rate.
example #2: After running a 10-K race Sheila drinks four cups of
water. Sheila enjoys each cup of water, but the first is the most
enjoyable, the second is a little less enjoyable, and so on. Sheila
gets enjoyment from consuming each cup of water, but the
 enjoyment declines a bit with each cup. Sheila’s utility is concave
(at least over 4 cups) in the number of cups of water that she
consumes (consuming too much water might eventually lead to
physical discomfort, a disutility, but that is another story).
 Formal Definition of Convexity and Concavity

 A function, f(X) is defined to be convex over some domain if for any

X 1 and X 2 in the domain of f(X), f[(X 1 + X 2 )/2] ≤ [f(X 1 ) + f(X 2 )]/2.
Furthermore, a function, f(X), is defined to be strictly convex over
some domain if for any X 1 and X 2 in the domain of f(X), f[(X 1 +
X 2 )/2] < [f(X 1 ) + f(X 2 )]/2.
 A function, f(X) is defined to be concave over some domain if for any
X 1 and X 2 in the domain of f(X), f[(X 1 + X 2 )/2] ≥ [f(X 1 ) + f(X 2 )]/2.
Furthermore, a function, f(X), is defined to be strictly concave over
some domain if for any X 1 and X 2 in the domain of f(X), f[(X 1 +
X 2 )/2] > [f(X 1 ) + f(X 2 )]/2.
examples:
y = x3
y = 0.5x0.5
y = 3x + 1

Variable Values

x 1 2 f(1+2)/2] [(f(1) +f(2)]/2

y = x3 1 8 3.375 4.5

y = 0.5x0.5 0.5 0.71 0.6124 0.6036

y = 3x + 1 4 7 5.5 5.5

 Comments
1. y = x3 is strictly convex since f(1+2)/2] < [(f(1) +f(2)]/2.
2. y = 0.5x0.5 is strictly concave since f(1+2)/2] > [(f(1) +f(2)]/2.
3. y = 3x + 1 is concave since f(1+2)/2] ≥ [(f(1) +f(2)]/2, but it is not
strictly concave. y = 3x + 1 is convex since f(1+2)/2] ≤ [(f(1)
+f(2)]/2, but it is not strictly convex. Functions that are both
concave and convex , f(1+2)/2] = [(f(1) +f(2)]/2, are said to be
affine.

REVIEW QUESTIONS

 Simplify the following equation to obtain an expression for y.

y - x = 3x + 4x -9

 What is the slope and the intercept of the following equations?

(1) y = 6x + 3
(2) y = -3x +9
(3) y = 4x

 Describe a convex function and a concave function using general terms.

 REVIEW QUESTIONS

 Simplify the following equation to obtain an expression for y

y - x = 3x + 4x -9
y - x + x = 3x + 4x + x - 9
y = x(3 + 4 + 1) - 9
then
y = 8x - 9

 What is the slope and the intercept of the following equations?

(1) slope = 6 and intercept = 3

(2) slope = -3 and intercept = 9
(3) slope = 4 and intercept = 0

 Describe a convex function and a concave function using general terms.

A convex function is a function the increases at an increasing rate.

A concave function is a function that increases at a decreasing rate.
II. BASIC STATISTICS

 LEARNING OBJECTIVES

 Understand a Basic Probability Distribution and Expectations

 Understand Measures of Central Tendency

 Arithmetic Mean
 Weighted Average
 Median
 Mode
 Geometric Mean

 Understand Measures of Dispersion

 Range
 Mean Absolute Deviation
 Standard Deviation
 Variance
 Coefficient of variation
 MEASURES OF CENTRAL TENDENCY

An measure of central tendency is a measure that shows some typical representation to a

group (either the population or a sample) of things or observations.

 Notation

 Index Notation
We will let X j represent any of the N values X 1 , X 2 , X 3 , …, X N for some
variable X. ‘j’ is the subscript or the index over which X is represented and
is usually noted j = 1, 2, 3, … , N.
 Summation Notation
The symbol Σ is the Greek letter sigma and is used in mathematics to
indicate summing over certain values.
example: Σ X j represents the summation of X j over j = 1, 2, 3, …, N
Σ X j = X 1 + X 2 + X 3 + …+ X N

 Probability Distribution
A set of possible outcomes of an events with a probability (likelihood) attached to
the occurrence (also called realization) of each outcome
example: Economy Return on Stocks Probability
Good 26% 0.25
Average 12% 0.50
Poor -2% 0.25
 Probability distributions may be discrete, such as the above example, or
continuous, such as the normal distribution (the ‘bell’ curve)

 Expectation (Expected Value)

The expectation (or expected value) is what we believe will be the outcome (e.g.,
the most likely or expected occurrence) of an uncertain event given a set of beliefs
about the future (a probability distribution). More formally we have:
If we let ‘p’ be some probability of an event or outcome X happening, the E(X) is
called the l expectation (or expected value) of X . For example, let X j represent
any of the N values X 1 , X 2 , X 3 , …, X N for some variable X. Let p j = p 1 , p 2 , …, p j ,
where Σ p j = 1. Then the expectation of X is equal to
E(X) = Σ p j X j = p 1 X 1 + p 2 X 2 + ,…, + p j X N .
 Expectation (Expected Value) -- continued
example: The expected value of the probability distribution in the last example is:
E(R) = 26%*0.25 + 12%*0.25 + (-2%)*0.25 = 12%
 Sometimes we denote the expected value of a variable, x, by E(x) = x^

 Arithmetic Mean

The arithmetic mean is often called the average

 The mean of a population, N (all the observations in the world), is called μ
N Xi X1 +X2 +...+XN-1 +XN
μ= ∑N = N
1=1

 The mean of a sample, n (not all the observations in the world), is called x-

n Xi X1 +X2 +...+Xn-1 +Xn

X = ∑ =
1=1 n n

example: Consider the following sample of students’ GMAT scores

620, 550, 590, 520, 480, 660, 510, 515

620+550+590+520+480+660+510+515
X = 8 = 555.625

 If a sample is unbiased (random) the sample median is known as an

unbiased estimator of the population mean
 If the sample is unbiased, the sample median is also an unbiased estimate of
the expected value of a ‘random draw’ of the population
 We often use the sample (or population) mean of historical data as a
forecast of the expected value of future events
example: The mean (average) return on the S&P 500 since 1929 is about 11%, so
we use 11% as our expected return on the S&P 500.
 Weighted Average
This is an average that incorporates some weighting mechanism for the
observations in question. If we assume that Σ w j = w 1 + w 2 + …+ w N = 1.0 then
Weighted μ = Σ w j X j = w 1 X 1 + w 2 X 2 + w 3 X 3 + …+ w N X N .
example: Grades in a class are determined as 25% midterm exam, 35% term paper,
and 40% final exam. A student makes 88 on the midterm, 94 on the term paper,
and 89 on the final.
Weighted Average = 88*.25 + 92*.35 + 89*.40 = 89.80 (Whew!)
 Note that our definition of an expectation is a weighted average where the
weights are probabilities.

 Median
The median is the value in a population (or sample) that has just as many
values above it as below it. If there are an even number of values, the
median will be the average of the two ‘middle’ values.
example: Consider our sample of student’s GMAT scores
620, 550, 590, 520, 480, 660, 510, 515
Arranging in descending order, we have
660,620,590,550,520,515,510,480
The median is (550 + 520)/2 = 535.

 Mode
The mode of a sample is the value that occurs most frequently or the greater number
of times. The mode does not have to be unique. For example, the sample 1, 2,
3, 3, 4, 4, 6 has two modes, 3 and 4 . A sample with two modes is called bimodal.

 Comparison of Mean, Median, and Mode

 The mean gives equal weighting to each observation. The median focuses
only on the observations in the middle of the sample distribution. The
mean uses the entire data set, but can be strongly influenced by extreme
values.
 There is only one mean and median, but potentially many modes.
 The mode is generally less useful as a measure of central tendency, but may
 sometimes be more useful in certain decision-making scenarios. For
example, knowing that the mean television screen sold was 26.3 inches is
not very useful in making an inventory decision. Knowing that the mode
was 25 inches is much more useful.

 Geometric Mean

The geometric mean, G, is calculated by taking the Nth root of the products a set of
N values or observations. For example, the geometric mean for a set of N values,
X 1 , X 2 , X 3 , …,X N is

G = N_ X 1 * X 2 * X 3 * …* X N = { X 1 * X 2 * X 3 * …* X N }1/N .
 We often use geometric means when we are calculating average returns
since the geometric return takes ‘compounding’ into account.
example: An investor earns the following returns: 12%, 15%, and 8% on an
investment over a 3 year period. The investor reinvests his gains each year.
Thus, the investor’s wealth increases by a factor of 1.12 in year 1, another 1.15 in
year 2, and 1.08 in year 3.
Geometric Mean = (1.12*1.15*1.08)_ = 1.1163. We subtract off 1 to take the
original investment into account, leaving us with a geometric return of 0.1163 or
11.63%. Note that the average return is Average = (12% + 15% + 8%)/3 =
11.67%.

 MEASURES OF DISPERSION

In addition to the central tendency of a population or sample, it is useful to have a measure

of its dispersion (how much the data varies from the mean). Consider the following
groups of numbers: (a) 3,5,7 and (b) 1,5,9. Both groups of numbers have the same mean,
5. It is apparent, however, that (b) is more disperse than (a). In this section, we discuss
measures which capture this dispersion

 Range

Dispersion or variation describes the degree to which values or observations spread

around some mean or average observation. The range for this set of values is the
difference between the largest or biggest value and the smallest.
example: Returning to our GMAT data, the largest value was 660 and the smallest
was 480. Thus, the range is 660 - 480 = 180.
 The range is useful in examining the high and the low, but fails to capture
dispersion in the middle of the sample. Consider the following groups of
numbers (a) 1,4,5,6,9 and (b) 1,3,5,7,9. Both groups have a mean of 5 and
a range of 8 (9-1). It is easy to see, however, that the two groups do not
 have the same dispersion: group ‘a’ is more closely clustered about the
mean than ‘b’.

 Mean Absolute Deviation (MAD)

 One candidate to capture the difference between a set of N values and their
mean is to take the difference from the mean and sum up the deviations.
Σ (X j - μ) = (X 1 - μ)+ (X 2 - μ)+ (X 3 - μ)+ …+ (X N - μ) .
The sum of deviations about the mean, however, will not provide any useful
information since it will be zero by construction.
example: 1,3,5,7,9; μ = 5.
(1-5) + (3-5) + (5-5) + (7-5) + (9-5) = -4 - 2 + 0 + 2 + 4 = 0!
 We get around this problem by taking the absolute value of each deviation
from the mean, summing them up, and taking the average (we take the
average because we are most interested in how a ‘typical’ observation
deviates from the mean. This measure is known as the Mean Absolute
Deviation (MAD).

N N
 ∑ (Xi -μ)  ∑ (Xi -X)
1=1 1=1
MAD = for a population MAD = N for a sample
N

example: 1,3,5,7,9; μ = 5.

(1-5)+(3-5)+(5-5)+(7-5)+(9-5)
MAD = = 2.4
5

 A MAD of 2.4 tells us that on average the absolute deviation from the mean
(ignoring algebraic signs) is 2.4.

 Variance and Standard Deviation

A more commonly used measure of dispersion is the variance, (denoted by σ2 for a
population and s2 for a sample) or the square root of the variance which is called the
standard deviation (denoted by σ or s respectively). The variance and standard
deviation are defined by mathematical equations which we will present a little later.
First, we note several important characteristics of variance and standard deviation.
 Unless all members of a population or a sample are the same number (e.g.,
 5,5,5...), the variance and standard deviation will never be zero.
 Adding or subtracting the same number to all observations will not change
the variance or standard deviation.
 The standard deviation is used with probability distributions to statistically
determine the proportion of data values on either side of the mean.
 Mathematical definition of variance

Variance is the average of the squared deviations from the mean:

N
(Xi -μ)2 n
(Xi-X)2
σ2 = ∑ N (Population) ∑ n-1 (Sample)
i=1 i=1

Comment: Using n-1 in the divisor for the sample variance is a technique that
provides a better estimate of the population variance in a sample. For larger
samples (e.g., n>30) the adjustment makes little difference.

example: Calculate the variance for the following sample:

10,15,12,18

First, note that the mean is

10+15+12+18
X = 4 = 13.75

Now, calculate the variance

2
(10-13.75)2+(15-13.75)2+(12-13.75)2+(18-13.75)2
s = 4-1 = 12.25

Comment: Note that squaring the deviations makes all deviations positive.
 Mathematical definition of standard deviation
Standard Deviation is the square root of Variance

σ= σ2 s = s2

example: The standard deviation of the last example is:

 s = 12.25 = 3.50

Comment: Note that the standard deviation is in the same units as the data (e.g.,
‘percent return’) while the variance is in the less intuitive ‘units squared’ (e.g., “percent
return squared”

 Coefficient of Variation

One problem with all of the preceding measures of dispersion is that they do
not control for scale effects. For instance, the standard deviation of A
={1000,1100,987} is 61.83 and the standard deviation of B ={1,20,50} is
24.70. Comparing 61.83 to to 24.70 might lead one to conclude that A is
more disperse than B. In absolute terms, this conclusion is true. In
percentage terms, however, it is not. Note that A has a mean of 1,029 and
B has a mean of 23.67. Relative to the mean, B deviates more than A!
We use coefficient of variation to measure dispersion relative to the mean.

σ s
CV = (Population) CV = X (Sample)
μ

example:

61.83 24.70
CVA = 1,029 = 0.0601 6.01% CVB = 23.67 = 1.0439 104.39%

Comment: In finance we often use standard deviation and coefficient of

variation as measures of risk -- the possibility that financial returns will
deviate from the expected value. In this sense, coefficient of variation
provides additional information that standard deviation does not since it
recognizes that most individuals will tolerate a greater dispersion if the
expected value is higher (a risk-return tradeoff).
 REVIEW QUESTIONS

 What is a probability distribution?

 Given the following data sample: {24,67,32,45}, calculate the following:

 Arithmetic Mean
 Median
 Range
 Variance
 Standard Deviation
 Coefficient of Variation

 Consider two samples, A and B. A has a mean of $1,000,000 and a standard

deviation of ($50,000). B has a mean of $500 and standard deviation of $200.
Which sample has the most dispersion? Justify your answer.
 REVIEW QUESTIONS -SOLUTIONS

 What is a probability distribution?

A probability distribution is a set of possible outcomes of an events with a probability

(likelihood) attached to the occurrence (also called realization) of each outcome. A
probability distribution may be discrete, or continuous.

 Given the following data sample: {24,67,32,45}, calculate the following:

 Arithmetic Mean
Mean = (24 + 67 + 32 + 45)/4 = 42.00
 Median
Ranking in descending order we get 67,45,32,24.
The median is (45 + 32)/2 = 38.5
 Range
The range = 67 - 24 = 43
 Variance
Variance = [(24-42)2 + (67-42)2 + (32-43)2 + (45-42)2]/3 = 352.6667
Note: divide by n-1 since this is a sample.
 Standard Deviation
Standard Deviation = _352.667 = 18.78
 Coefficient of Variation
CV = 18.78/42 = 0.4471 or 44.71%

 Consider two samples, A and B. A has a mean of $1,000,000 and a standard

deviation of ($50,000). B has a mean of $500 and standard deviation of $200.
Which sample has the most dispersion? Justify your answer.
In absolute terms, A has more dispersion since it has a much larger standard
deviation. In percentage terms, however, the CV of A is 0.05 while the CV of B is
0.40. Thus, B is much more disperse relative to the mean.