QUANTITATIVE METHODS: APPLICATIONS

COMMON PROBABILITY DISTRIBUTIONS

 DISCLAIMER
CFA INSTITUTE DOES NOT ENDORSE, PROMOTE, REVIEW,
OR WARRANT THE ACCURACY OF THE PREPARATORY
SOURCES OFFERED BY LOMONOSOV MOSCOW STATE
UNIVERSITY OR VERIFY OR ENDORSE THE PASS RATES
CLAIMED BY LOMONOSOV MOSCOW STATE UNIVERSITY.

CFA®, AND CHARTERED FINANCIAL ANALYST® ARE

TRADEMARKS OWNED BY CFA INSTITUTE.

READING 10 COMMON PROBABILITY DISTRIBUTIONS 2

 A PROBABILITY DISTRIBUTION
DISCRETE AND CONTINUOUS RANDOM VARIABLES AND THEIR PROBABILITY FUNCTIONS

 Probability determines the likelihood of some event – bounded by 0,1

 Tossing a fair coin – probability of getting a head is equal to one-half.

 Probability distribution maps each set of outcomes (head/tail) to its probability.

 Discrete random variable – the number of possible values is finite (tossing a

coin).

 Continuous random variable – the number of possible values is infinite (variable

that takes values between 0,1 ).

READING 10 COMMON PROBABILITY DISTRIBUTIONS 3

 QUESTION

What is the probability that random variable that takes values between 0,1
equals 0.5:

A) 0.0

B) 1.0

C) 0.5

Since the number of possible outcomes is infinite, the only possible probability is
zero.

READING 10 COMMON PROBABILITY DISTRIBUTIONS 4

 THE SET OF POSSIBLE OUTCOMES OF A SPECIFIED DISCRETE RANDOM VARIABLE.

 Tossing a fair coin with two possible outcomes – Head and Tail.

 Since the coin is fair, probabilities should be equal - 𝑃 𝐻 = 𝑃 𝑇 = 1 2.

 The sum of probabilities of all the possible outcomes should be equal to 1.

READING 10 COMMON PROBABILITY DISTRIBUTIONS 5

 QUESTION

Throwing a fair 6-sided dice. What is the probability of getting 4?

A) 0.0

B) 1 6

C) 1 4

Since the dice is fair, the probability should be equal to one over six.

READING 10 COMMON PROBABILITY DISTRIBUTIONS 6

 CUMULATIVE DISTRIBUTION FUNCTION

 Cumulative Distribution Function (CDF) - probability that random variable is

less than or equal to a given value: 𝐹 𝑥 = 𝑃 𝑋 ≤ 𝑥

 Again consider a coin flipping experiment: define the random variable 𝑋 that is
equal to one if we flip a head, and zero if we have a tail.

 Clearly 𝑃 𝐻 = 𝑃 𝑇 = 1 2

CDF 𝑋

1.0

0.5

0 1 𝑥

READING 10 COMMON PROBABILITY DISTRIBUTIONS 7

CALCULATION AND INTERPRETATION OF PROBABILITIES FOR A RANDOM VARIABLE,
GIVEN ITS CUMULATIVE DISTRIBUTION FUNCTION.

For a discrete random variable a CDF can be written as a sum of appropriate

probabilities

𝐹 𝑥 =𝑃 𝑋≤𝑥 = 𝑃 𝑋 = 𝑥𝑖
𝑖:𝑥𝑖 ≤𝑥

READING 10 COMMON PROBABILITY DISTRIBUTIONS 8

 QUESTION

Throwing a fair 6-sided dice. What is the probability of getting less than or
equal to 4?

A) 1 3

B) 1 6

C) 2 3

Since the dice is fair, the probability of each outcome should be equal to one over
six. Sum them up to get CDF.
READING 10 COMMON PROBABILITY DISTRIBUTIONS 9
 A DISCRETE UNIFORM RANDOM VARIABLE,
A BERNOULLI RANDOM VARIABLE, AND A BINOMIAL RANDOM VARIABLE.

 Discrete uniform random variable is a discrete random variable with equally

likely outcomes (throwing a 6-sided die).

 Bernoulli random variable is a discrete random variable with only two

outcomes (tossing a coin).

 Binomial random variable is a discrete random variable that counts the number
of “successes” in a series of Bernoulli trials. Fix the number of trials 𝑛 , we are
interested in the probability that the number of “successes” (number of heads
during tossing, probability of a success in each trial is 𝑝) is equal to 𝑘:

𝑛!
P 𝑋=𝑘 = 𝑝𝑘 1 − 𝑝 𝑛−𝑘
𝑘! 𝑛 − 𝑘 !

READING 10 COMMON PROBABILITY DISTRIBUTIONS 10

CALCULATION AND INTERPRETATION PROBABILITIES GIVEN THE DISCRETE UNIFORM
AND THE BINOMIAL DISTRIBUTION FUNCTIONS

For a uniform random variable, the probabilities for every probable outcome are
equal:

P 𝑋 = 𝑥1 = ⋯ = P 𝑋 = 𝑥𝑛

READING 10 COMMON PROBABILITY DISTRIBUTIONS 11

 QUESTION

Calculate the probability that the uniform random variable (𝑋 = 1,2,3,4 ) takes
value between 2 and 3 - 𝑃 2 ≤ 𝑋 ≤ 3

A) 1 2

B) 1 4

C) 1 3

As a first step calculate the probability of each outcome – 1/4. Since only two out of
four possible outcomes lie in the interval we should multiply 1/4 by two.
READING 10 COMMON PROBABILITY DISTRIBUTIONS 12
 QUESTION

Calculate the probability that the binomial random variable

(𝑛 = 5, 𝑝 = 0.25) takes value between 2 and 3 - 𝑃 2 ≤ 𝑋 ≤ 3

A) 0.60

B) 0.35

C) 0.15

We should sum up probabilities of all the outcomes lying in the interval. Hence,
𝑃 2≤𝑋 ≤3 =𝑃 𝑋 =2 +𝑃 𝑋 =3 :
𝑃 2≤𝑋≤3
5! 5!
= 0.252 1 − 0.25 5−2 + 0.253 1 − 0.25 5−3
2! 5 − 2 ! 3! 5 − 3 !

READING 10 COMMON PROBABILITY DISTRIBUTIONS 13

 DESCRIBING STOCK PRICE MOVEMENT WITH BINOMIAL TREE

 Fix the value of stock today 𝑆0 . Suppose our stock price can be only in two
states over the next period 𝑆1 .

 Probability of an up move 𝑆1 = 𝑢𝑆0 , 𝑢 > 1 equals to 𝑝, and 1 − 𝑝 for a down

move 𝑆1 = 𝑑𝑆0 , 𝑑 < 1 . Additional condition - 𝑢𝑑 = 1

𝑢𝑢𝑢𝑆0
𝑢𝑢𝑆0

𝑢𝑆0
𝑢𝑆0

𝑆0 𝑆0 Number of possible paths

to reach this node is
𝑑𝑆0 higher than the number
𝑑𝑆0
of paths to reach the
𝑑𝑑𝑆0 upper node.

𝑑𝑑𝑑𝑆0

READING 10 COMMON PROBABILITY DISTRIBUTIONS 14

 QUESTION

Calculate the stock value at the chosen node (see the tree plot). Calculate the
probability of getting there (𝑆0 = 100, 𝑝 = 0.4, 𝑢 = 1.1, 𝑑 = 1/1.1).

A) 110; 0.37

B) 110; 0.29

C) 91; 0.29

For this node 𝑆1 = 𝑢𝑆0 = 110. The probability can be calculated using binomial
distribution:
3!
𝑃 𝑆1 = 𝑢𝑆0 = 0.61 1 − 0.6 3−1
1! 3 − 1 !

READING 10 COMMON PROBABILITY DISTRIBUTIONS 15

 TRACKING ERROR

Tracking error is a measure of how closely a portfolio matches the benchmark.

For example our portfolio is constructed to follow S&P 500 Index. If S&P 500 index
gained 6% over the last year, and our portfolio increased by 10%. The tracking error
should be 10% - 6%= 4%.

READING 10 COMMON PROBABILITY DISTRIBUTIONS 16

 THE CONTINUOUS UNIFORM DISTRIBUTION
CALCULATION PROBABILITIES, GIVEN A CONTINUOUS UNIFORM DISTRIBUTION

 Since for a continuous variable the number of possible outcomes is infinite, the
probability of taking any particular value is zero, and we should proceed with
another approach of determining a CDF (in contrast to discrete).
 The most common example of a continuous random variable – uniform
distributed with maximum value of 1 and minimum of 0 (all values between are
equally likely). Denoted as 𝑋~𝑈 0,1

 One should memorize the following formula for a 𝑈 𝑎, 𝑏 CDF:

𝐹 𝑥 = 0, 𝑥 ≤ 𝑎

𝑥−𝑎
𝐹 𝑥 = ,𝑎 ≤ 𝑥 < 𝑏
𝑏−𝑎

𝐹 𝑥 = 1, 𝑏 ≤ 𝑥

READING 10 COMMON PROBABILITY DISTRIBUTIONS 17

 QUESTION

Let 𝑋~𝑈 3,7 , what is the probability of lying within the interval 4,5

A) 0.5

B) 0.75

C) 0.25

The distance between 4 and 5 is one, that is four times less than the distance
between 3 and 7.

READING 10 COMMON PROBABILITY DISTRIBUTIONS 18

 THE KEY PROPERTIES OF THE NORMAL DISTRIBUTION

 Normal distribution - denoted 𝑁 𝜇, 𝜎 2 , is the most widely used distribution in

finance.
 The key properties:
1) 𝜇 – mean, distribution parameter, center measure
2) 𝜎 – standard deviation, distribution parameter, spread measure
3) kurtosis, equals to 3, tailedness measure
4) skewness, equals to 0, symmetricity measure

 Density plot (“likelihood”):

READING 10 COMMON PROBABILITY DISTRIBUTIONS 19

 QUESTION

Which of the following set of parameters corresponds to the normal distribution

with the highest variance:

A) Skewness = 0.0; 𝜎 = 0.4

B) Skewness = 0.2; 𝜎 = 0.8

C) Skewness = 0.0; 𝜎 = 0.7

Normal distribution is symmetrical. Variance - 𝜎 2

READING 10 COMMON PROBABILITY DISTRIBUTIONS 20

 A UNIVARIATE AND A MULTIVARIATE DISTRIBUTION
THE ROLE OF CORRELATION IN THE MULTIVARIATE NORMAL DISTRIBUTION

 Multivariate distribution is a joint distribution of random variables.

 Multivariate normal distribution of 𝑁 random variables can be specified by 𝑁

means, 𝑁 standard deviations, and 𝑁 ∙ (𝑁 − 1)/2 pairwise correlations.

 Correlation “defines” the measure of comovement.

READING 10 COMMON PROBABILITY DISTRIBUTIONS 21

 DETERMINING THE PROBABILITY THAT A NORMALLY DISTRIBUTED RANDOM
VARIABLE LIES INSIDE A GIVEN INTERVAL

 Let 𝑋~𝑁 𝜇, 𝜎 2 . The following facts should be learned by heart:

The probability that 𝑋 lies in 𝜇 − 1.65𝜎, 𝜇 + 1.65𝜎 - 90%
The probability that 𝑋 lies in 𝜇 − 1.96𝜎, 𝜇 + 1.96𝜎 - 95%
The probability that 𝑋 lies in 𝜇 − 2.58𝜎, 𝜇 + 2.58𝜎 - 99%

 𝜇 and 𝜎 can be estimated empirically.

READING 10 COMMON PROBABILITY DISTRIBUTIONS 22

 QUESTION

Which of the following intervals corresponds to the 95% confidence interval for a
normal variable 𝑋~𝑁 1,9 :

A) −4.88,6.88

B) −3.95,5.95

C) −2.35,3.26

1 ± 1.96 ∙ 3

READING 10 COMMON PROBABILITY DISTRIBUTIONS 23

 THE STANDARD NORMAL DISTRIBUTION
STANDARDIZATION OF A RANDOM VARIABLE
CALCULATION PROBABILITIES USING THE STANDARD NORMAL DISTRIBUTION

 Standard normal variable – normal variable with a mean zero and standard
deviation equal to one 𝑋~𝑁 0,1
 To convert any normal variable 𝑋~𝑁 𝜇, 𝜎 2 to a standard normal variable
𝑌~𝑁 0,1 (standardize it) one should subtract mean and divide by standard
normal deviation:
𝑋−𝜇
𝑌=
𝜎

 By plugging an actual 𝑋 value into the formula caclulate z-value.

 Since any normal variable can be standardized, it is enough to know the CDF for
a standard one.

READING 10 COMMON PROBABILITY DISTRIBUTIONS 24

 QUESTION

What is the probability that 𝑋~𝑁 3,4 will be between 2 and 4:

A) 𝑁 0.5 − 𝑁 −0.5

B) 𝑁 −0.5 + 𝑁 0.5

C) 𝑁 0.5 − 𝑁 0.25

Step 0. Determine what do we need to calculate - R = 𝑃 𝑋 ≤ 4 − 𝑃 𝑋 ≤ 2

Step 1. Standardize 𝑋.
𝑋−𝜇 4−3 2−3
𝑌= → 𝑋 = 𝜇 + 𝜎 ∙ 𝑌 and R = 𝑃 𝑌 ≤ −𝑃 𝑌 ≤
𝜎 2 2
Step 2. Look up in a given z-table values for the standard normal CDF 𝑁 at points
-0.5 and 0.5

Pay attention: 𝑁 𝑥 = 1 − 𝑁(−𝑥)

READING 10 COMMON PROBABILITY DISTRIBUTIONS 25

 THE SHORTFALL RISK, THE SAFETY-FIRST RATIO, AND
SELECTING AN OPTIMAL PORTFOLIO USING ROY’S SAFETY-FIRST CRITERION

 Shortfall risk – risk for the asset to slump below some stated level (e.g. risk that
CO1 loses more than 10% over the next year)
 Safety-first ratio equals the difference between our portfolio expected return
and the barrier return (-10% in the example above) divided by the standard
deviation of portfolio returns:
𝐸 𝑅 − 𝑅𝐵
𝑆𝐹𝑅 =
𝜎

 In order to choose a portfolio with the minimal shortfall risk, we should pick the
portfolio with the highest SFR – Roy’s safety first criterion.

READING 10 COMMON PROBABILITY DISTRIBUTIONS 26

 THE RELATIONSHIP BETWEEN NORMAL AND LOGNORMAL DISTRIBUTIONS
THE USE OF THE LOGNORMAL DISTRIBUTION TO MODEL ASSET PRICES

 Random variable 𝑋 is said to follow a lognormal distribution if its logarithm

follows a normal distribution.
 Natural exponential function is always greater than zero, hence a lognormally
distributed variable is bounded by zero (the majority of assets cannot have
negative prices).
 Pay attention to the skewness on the density plot below:

READING 10 COMMON PROBABILITY DISTRIBUTIONS 27

 DISCRETELY AND CONTINUOUSLY COMPOUNDED RATES OF RETURN
CALCULATION OF A CONTINUOUSLY COMPOUNDED RATE OF RETURN, GIVEN A
SPECIFIC HOLDING PERIOD RETURN

 Consider discretely compounded rates. Let 𝑚 be the number of compounding

periods in a year, and 𝑟 be the stated rate. The effective annual rate can be
calculated as:
𝑟 𝑚
𝐸𝐴𝑅 = 1 + −1
𝑚
 As 𝑚 goes to infinity, 𝐸𝐴𝑅 goes to the limit:

𝐸𝐴𝑅 = 𝑒 𝑟𝑐 − 1
 rc is known as a continuous compounded rate.
 Holding period return shows the return of an asset over a specific period (𝑋0 −
initial value, 𝑋1 - value at the end of period):
𝑋1
𝑟𝑐 = ln = ln 1 + 𝐻𝑃𝑅
𝑋0

READING 10 COMMON PROBABILITY DISTRIBUTIONS 28

 QUESTION

Calculate the EAR for a quarterly compounding, with a rate of 12%:

A) 13.23

B) 11.42

C) 12.55

4
12%
𝐸𝐴𝑅 = 1 + −1
4
READING 10 COMMON PROBABILITY DISTRIBUTIONS 29
 MONTE CARLO SIMULATION AND ITS APPLICATIONS AND LIMITATIONS

 Monte Carlo methods are based on the repeating simulation of some random
process (tossing a coin, simulating asset value) to get the final result via
averaging.

 For example we tossed a coin 1000 times – 501 heads and 499 tail. Hence our
501
estimate of the probability of a head is
1000

 Methods are widely used in derivatives pricing and risk management.

 The main limitation is complexity:

• If a random process is incorrectly specified the method will not converge;

• Large number of paths is usually needed to reach the required accuracy.

READING 10 COMMON PROBABILITY DISTRIBUTIONS 30

 MONTE CARLO SIMULATION AND HISTORICAL SIMULATION

 Monte Carlo methods simulate a random process based on the specified

assumptions (e.g. final stock distribution is lognormal).

 Historical method performs simulations using only the historical data (e.g. we
know the stock returns for the last 100 days) assuming that the future
distribution is the same.

 The most crucial limitation of the historical simulation method – the history can
be not representative.

READING 10 COMMON PROBABILITY DISTRIBUTIONS 31

 HOMEWORK ASSIGNMENT
READING
CFA® Level I Curriculum (2019) Volume I  Reading 10

PRACTICE PROBLEMS
CFA® Level I Curriculum (2019) Volume I  Reading 10  PRACTICE PROBLEMS
MOODLE  CFA® Level I 2019  TESTS  QM #3

READING 10 COMMON PROBABILITY DISTRIBUTIONS 32