Risk Management
Lecture 2-1: Probability Theory
Chen Tong
SOE & WISE, Xiamen University
September 12, 2024
Chen Tong (SOE&WISE) Risk Management September 12, 2024 1 / 62
Probability Theory
This chapter includes:
▶ Random Variables and Probability Distributions
▶ Joint Distributions
▶ Features of Probability Distributions
▶ The Normal and Related Distributions
Chen Tong (SOE&WISE) Risk Management September 12, 2024 2 / 62
Probability Theory
1. Random Variables and Probability Distributions
Chen Tong (SOE&WISE) Risk Management September 12, 2024 3 / 62
▶ Experiment: flip a coin and count the number of times the coin
turns up heads.
▶ In theory, the experiment can be indefinitely repeated and has a
well-defined set of outcomes.
▶ A random variable takes on numerical values and has an outcome
that is determined by the experiment.
▶ Notation: random variables are usually denoted by upper case letters
(e.g. X), particular realizations are denoted by the corresponding
lowercase letters (e.g. x = 3 ).
▶ Example: X = 1 if the coin turns up heads and X = 0 if the coin
turns up tails.
▶ Random variables that only take on the values zero and one are
called Bernoulli variables or binary variables.
Chen Tong (SOE&WISE) Risk Management September 12, 2024 4 / 62
▶ We often consider discrete random variables (0, 1, 2, . . .).
▶ Simplest example: Bernoulli variable:
P(X = 1) = 0.5, P(X = 0) = 0.5, P(X = 1) + P(X = 0) = 1
▶ In general, the probability can be any number between zero and one.
Call this number θ :
P(X = 1) = θ, P(X = 0) = 1 − θ
▶ If X takes on k possible values {x1 , x2 , . . . , xk }, then the
probabilities p1 , p2 , . . . , pk are defined by
X
pj = P (X = xj ) , j = 1, 2, . . . , k, pj = 1
j
▶ The probability density function (pdf) of X summarizes the
information concerning the possible outcomes of X and the
corresponding probabilities:
pj = f (xj ) , j = 1, 2, . . . , k
Chen Tong (SOE&WISE) Risk Management September 12, 2024 5 / 62
▶ Example: number of free throws made by a basketball player out of
two attempts, {0, 1, 2}
▶ pdf of X is given by f (0) = .20, f (1) = .44, f (2) = .36
Chen Tong (SOE&WISE) Risk Management September 12, 2024 6 / 62
▶ The probability to observe a certain realization xj is always between
zero and one:
0 ≤ P (X = xj ) ≤ 1
▶ The sum of the probabilities of all realizations of a random variable
is always equal to one:
X k
f (xj ) = 1
j=1
▶ The cumulative distribution function (cdf) of a discrete random
variable is obtained by summing the pdf over all values xj such that
xj ≤ x( for a given point x)
X
F (x) ≡ f (x) = P(X ≤ x)
xj ≤x
Chen Tong (SOE&WISE) Risk Management September 12, 2024 7 / 62
▶ cdf of a discrete random variable
Chen Tong (SOE&WISE) Risk Management September 12, 2024 8 / 62
▶ Continuous random variables take on real values
▶ It makes no sense to calculate the probability that a continuous
random variable takes on a particular value
▶ We compute events involving a range of values:
▶ Example: constants a and b with a < b
▶ The probability that X lies between the numbers a and b,
P(a ≤ X ≤ b), is the area under the pdf between points a and b
Chen Tong (SOE&WISE) Risk Management September 12, 2024 9 / 62
▶ Continuous random variables may be described by a pdf which is
defined as a non-negative function for any real number x,
Z +∞
f (x) ≥ 0 so that f (x)dx = 1
−∞
▶ The probability of a realization in the interval [a, b] is the area under
the pdf from a to b,
Z b
P(a ≤ X ≤ b) = f (x)dx ≥ 0
a
▶ The cdf of a continuous random variable is defined as
Z x
∆F (x)
F (x) ≡ f (x)dx = P(X ≤ x) with f (x) =
−∞ ∆x
Chen Tong (SOE&WISE) Risk Management September 12, 2024 10 / 62
▶ cdf of a continuous random variable
Chen Tong (SOE&WISE) Risk Management September 12, 2024 11 / 62
Properties:
▶ 0 ≤ F (x) ≤ 1
▶ if x2 > x1 , then F (x2 ) ≥ F (x1 )
▶ F (+∞) = 1
▶ F (−∞) = 0
▶ For any number c, P(X > c) = 1 − F (c)
▶ For any numbers a < b, P(a < X ≤ b) = F (b) − F (a)
Chen Tong (SOE&WISE) Risk Management September 12, 2024 12 / 62
Probability Theory
2. Joint Distributions
Chen Tong (SOE&WISE) Risk Management September 12, 2024 13 / 62
▶ We are usually interested in more than one variable
⇒ Joint distributions
▶ Consider two random variables X and Y with a joint probability
distribution which can be described by the joint probability density
function fX ,Y (x, y )
▶ If X and Y are discrete random variables, then the joint probability
density function is given by
fX ,Y (x, y ) = P(X = x, Y = y )
▶ Properties:
P P
fX ,Y (x, y ) ≥ 0, X y fX ,Y (x, y ) = 1 in discrete case
R R
fX ,Y (x, y ) ≥ 0, X y fX ,Y (x, y )dydx = 1 in continuous case
Chen Tong (SOE&WISE) Risk Management September 12, 2024 14 / 62
Example:
21 2
if x 2 ≤ y ≤ 1
fX ,Y (x, y ) = 4 x y
0 otherwise
Since x 2 ≤ y ≤ 1 and −1 ≤ x ≤ 1, fX ,Y (x, y ) cannot be negative:
fX ,Y (x, y ) ≥ 0
The area under the joint distribution is equal to one:
Z 1 Z 1 Z 1 Z 1 Z 1 1 !
21 2 21 2 2
fX ,Y (x, y )dydx = x y dydx = x y dx
−1 x2 −1 x2 4 −1 8 x2
Z 1
21 2 21 2 4
= x − x x dx
−1 8 8
Z 1
21 2 4
= x 1−x dx
−1 8
1
21 3 21 7 4 4
= x − x = + =1
24 56 −1 8 8
Chen Tong (SOE&WISE) Risk Management September 12, 2024 15 / 62
▶ The realization of X and Y within a given interval is
Z b Z d
P(a ≤ X ≤ b, c ≤ Y ≤ d) = fX ,Y (x, y )dydx
a c
The cumulative joint probability density functions
F (z, w ) = P(X ≤ z, Y ≤ w )
are XX
F (z, w ) = fX ,Y (x, y )
x≤z y ≤w
in the discrete case and
Z z Z w
F (z, w ) = fX ,Y (x, y )dydx
−∞ −∞
in the continuous case
Chen Tong (SOE&WISE) Risk Management September 12, 2024 16 / 62
Marginal pdf
▶ If X and Y are independent,
fX ,Y (x, y ) = fX (x)fY (y ),
where fX (x) and fY (y ) are marginal probability density functions,
defined as:
P
fX (x) = R y fX ,Y (x, y ) in the dicrete case
f (x, s)ds in the continuous case
y X ,Y
and
P
fY (y ) = R X fX ,Y (x, y ) in the dicrete case
f (s, y )ds in the continuous case
X X ,Y
Example above:
Z 1 1
21 2 21 2 2 21 2 21 2 4 21 2
x 1 − x4
fX (x) = x ydy = x y = x − x x =
x2 4 8 x2 8 8 8
Chen Tong (SOE&WISE) Risk Management September 12, 2024 17 / 62
Conditional Distributions (∗ ∗ ∗)
▶ In econometrics, we are interested in how a random variable X (or a
set of random variables) affects a random variable Y
⇒ Conditional probability density function of Y given X :
fX ,Y (x, y )
fY |X (y | x) = with fX (x) > 0
fX (x)
▶ Properties: fY |X (y | x) ≥ 0 and
P
R ∞y fY |X (y | x) = 1 in the discrete case
f (y | x)dy = 1 in the continuous case
−∞ Y |X
▶ Relationship between conditional and joint probabilities:
fY ,X (y , x) = fY |X (y | x)fX (x)
21 2
▶ Example above: fY |X (y | x) = f (x,y ) 4 x y 2y
fX (x) = 21 2 4 = 1−x 4
8 x (1−x )
Chen Tong (SOE&WISE) Risk Management September 12, 2024 18 / 62
Copula Function
▶ Sklar’s theorem states that any multivariate joint distribution can be
written in terms of univariate marginal distribution functions and a
copula which describes the dependence structure between the
variables.
▶ In bivariate case
f (x, y ) =f (x) · f (y ) · C (F (x), F (y )) ,
or expressed it in log-form
log f (x, y ) = log f (x) + log f (y ) + log C (F (x), F (y )) ,
where f (x) and f (y ) are the marginal pdf, and F (x) and F (y ) are
the marginal cdf.
Chen Tong (SOE&WISE) Risk Management September 12, 2024 19 / 62
Gaussian Copula Function
▶ For two (standardized) random variable x and y , a commonly used
copula function is
1 1 −1
x
C (x, y ) = p exp − (x, y ) R − I2
|R| 2 y
where R is the correlation matrix of x and y , and I is the identity
matrix.
▶ David Li (2000) "Gaussian Copula Model for Pricing CDO"
(collateralized debt obligations):
The Formula That Killed Wall Street.
▶ A more detailed introduction to Copula function will be given in
later class.
Chen Tong (SOE&WISE) Risk Management September 12, 2024 20 / 62
Probability Theory
3. Features of Probability Distributions
Chen Tong (SOE&WISE) Risk Management September 12, 2024 21 / 62
▶ We are mainly interested in a few aspects of the distributions of
random variables:
▶ "Measures of central tendency":
▶ Expected value (weighted average of all possible values of X )
▶ Median (splits the pdf into two equal parts)
▶ "Measures of variability":
▶ Variance (squared difference of X from its expected value)
▶ Standard deviation (square root of the variance)
▶ "Measures of association between two random variables":
▶ Covariance (measure of linear dependence between two random
variables)
▶ Correlation coefficient (unit-free measure of linear dependence
between two random variables)
Chen Tong (SOE&WISE) Risk Management September 12, 2024 22 / 62
Expected Value
▶ The expected value (or mean) E (X ) = µX of a discrete random
variable X is the weighted average of all possible realizations of X ,
where the probabilities of the realizations x are used as weights:
X
E (X ) = µX = xf (x)
x
▶ For a continuous random variable X , the expected value is
Z ∞
E (X ) = µX = xf (x)dx
−∞
▶ Calculation rules:
E (a) = a,
E (X + Y ) = E (X ) + E (Y ) = µX + µY
E (aX + b) = aE (X ) + b = aµX + b
Chen Tong (SOE&WISE) Risk Management September 12, 2024 23 / 62
Variance
▶ We do not only want to know the expected value but also the spread
of a distribution
⇒ How far is the distance of a random variable X from its expected
value?
▶ We usually consider the squared difference (X − µX )2
⇒ Elimination of signs and stronger "punishment" of larger
distances
Chen Tong (SOE&WISE) Risk Management September 12, 2024 24 / 62
▶ Since (X − µX )2 is a random variable itself, we may calculate the
expected distance from X to its mean:
( P
2
h
2
i f (x) (x − µX ) if discrete
Var(X ) = σX2 ≡ E (X − µX ) = R ∞X 2
−∞
f (x) (x − µX ) dx if continuous
▶ Calculation rule:
Var(aX + b) = Var(aX ) = a2 Var(X ) = a2 σX2
▶ The standard deviation sd(X ) of a random variable X p
is the
(positive) square root of the variance: sd(X ) = σX ≡ Var(X )
Chen Tong (SOE&WISE) Risk Management September 12, 2024 25 / 62
Chen Tong (SOE&WISE) Risk Management September 12, 2024 26 / 62
Standardizing a Random Variable
▶ We may define a new random variable Z by substracting its mean
and dividing by its standard deviation:
X −µ
Z=
σ
rewritten as Z = aX + b with a ≡ (1/σ) and b ≡ −(µ/σ)
⇒
E (Z ) = E (aX + b) = aE (X ) + b = (µ/σ) − (µ/σ) = 0
and
Var(Z ) = Var(aX + b) = a2 Var(X ) = σ 2 /σ 2 = 1
▶ Z is called a standardized random variable
Chen Tong (SOE&WISE) Risk Management September 12, 2024 27 / 62
Moments
▶ The nth central moment of the probability distribution of a random
variable X is n
µn = E (X − µX )
h i
▶ The first central moment is E (X − µX )1 = E (X ) − µX = 0
h i
▶ The second central moment is the variance: E (X − µX )2
Chen Tong (SOE&WISE) Risk Management September 12, 2024 28 / 62
h i
▶ The third central moment E (X − µX )3 measures the skewness of
the distribution
▶ The third central moment of a symmetric distribution is zero
▶ If the distribution is skewed to the left, it has a negative skewness
▶ If the distribution is skewed to the right, it has a positive skewness
(example: wage distribution)
" 3 #
X −µ
Skew(X ) = E
σ
▶ The fourth central moment (kurtosis) is larger if the tails in the
distribution of X are thicker
" 4 #
X −µ
Kurt(X ) = E
σ
Chen Tong (SOE&WISE) Risk Management September 12, 2024 29 / 62
Covariance
▶ The covariance is a measure of the linear relationship between two
random variables X and Y
▶ Consider the random variable (X − µX ) (Y − µY ) :
▶ The covariance is the expected value of (X − µX ) (Y − µY ) :
Cov(X , Y ) = σXY ≡ E [(X − µX ) (Y − µY )]
P P
= R ∞x R y∞f (x, y ) (x − µX ) (y − µY ) if discrete
−∞ −∞
f (x, y ) (x − µX ) (y − µY ) dxdy if continuous
Chen Tong (SOE&WISE) Risk Management September 12, 2024 30 / 62
Covariance
▶
Cov(X , Y ) = E [(X − µX ) (Y − µY )]
= E [XY − µX Y − µY X + µX µY ]
= E (XY ) − µX µY
▶ If X and Y are independent, then their covariance is zero:
E (XY ) = E (X )E (Y ) = µX µY ⇒ Cov(X , Y ) = 0
- However, the converse is not necessarily true, i.e. random variables
may have a covariance of zero although they are not independent
Chen Tong (SOE&WISE) Risk Management September 12, 2024 31 / 62
Covariance
▶ Any random variable with E (X ) = 0 and E X 3 = 0 has the
property that, if Y = X 2 , then Cov(X , Y ) = 0
▶ X and Y = X 2 are clearly not independent
⇒ Weakness of the covariance as a general measure of association
between random variables
⇒ The covariance is useful in contexts when relationships are at
least approximately linear
▶ Calculation rules for variances and covariances:
Cov(X , X ) = Var(X ) Cov(aX + b, cY + d) = acCov(X , Y )
Var(X + Y ) = Var(X ) + Var(Y ) + 2 Cov(X , Y ),
Var(aX + bY ) = a2 Var(X ) + b 2 Var(Y ) + 2ab Cov(X , Y )
Chen Tong (SOE&WISE) Risk Management September 12, 2024 32 / 62
Correlation Coefficient
▶ The fact that the covariance depends on units of measurement is a
deficiency that is overcome by the correlation coefficient between X
and Y :
Cov(X , Y ) σXY
Corr(X , Y ) = ρXY = =
sd(X )sd(Y ) σX σY
▶ Cauchy-Schwarz inequality: | Cov(X , Y )| ≤ sd(X ) sd(Y )
−1 ≤ Corr(X , Y ) ≤ 1
▶ Cov(X , Y ) and Corr(X , Y ) always have the same sign
Chen Tong (SOE&WISE) Risk Management September 12, 2024 33 / 62
Correlation Coefficient
▶ If X and Y are independent, then Corr(X , Y ) = 0, but zero
correlation does not imply independence
▶ The correlation between X and Y is invariant to the units of
measurement of X or Y :
▶ If a1 a2 > 0 ⇒ Corr (a1 X + b1 , a2 Y + b2 ) = Corr(X , Y )
▶ If a1 a2 < 0 ⇒ Corr (a1 X + b1 , a2 Y + b2 ) = − Corr(X , Y )
Chen Tong (SOE&WISE) Risk Management September 12, 2024 34 / 62
Conditional Expectation
▶ Covariance and correlation measure the linear relationship between
two random variables and treat them symmetrically. We usually
want to explain Y in terms of X
▶ Suppose we know that X has taken on a particular value x
⇒ We can compute this expected value of Y , given that we know
this outcome of X
▶ We denote the expected value by E (Y | X = x)
Chen Tong (SOE&WISE) Risk Management September 12, 2024 35 / 62
The Law of Iterated Expectations
▶ When Y is a discrete random variable taking on values {y1 , . . . , ym },
then
Xm
E (Y | X = x) = yj fY |X (yj | x)
j=1
▶ When Y is continuous, then
Z +∞
E (Y | X = x) = yfY |X (y | x)dy
−∞
▶ E (Y | X = x) tells us how the expected value of Y varies with x
Chen Tong (SOE&WISE) Risk Management September 12, 2024 36 / 62
Chen Tong (SOE&WISE) Risk Management September 12, 2024 37 / 62
▶ Given the random variables Y and X , the conditional expected value
E (Y | X ) is a random variable whose value depends on the value of
X
"Note that E (Y | X = x) is a function of x. If
E (Y | X = x) = g (x), then E (Y | X ) = g (X )"
▶ The law of iterated expectations states that the expected value of
the conditional expected value of Y given X equals the expected
value of Y :
E [E (Y | X )] = E (Y )
Chen Tong (SOE&WISE) Risk Management September 12, 2024 38 / 62
▶ Proof for discrete random variables:
X
E [E (Y | X )] = E (Y | X = x)P(X = x)
x
XX
= yP(Y = y | X = x)P(X = x)
x y
XX
= yP(Y = y , X = x) = E (Y )
x y
Chen Tong (SOE&WISE) Risk Management September 12, 2024 39 / 62
Independence and Correlation
▶ If X and Y are independent, then E (Y | X ) = E (Y )
▶ If E (Y | X ) = E (Y ), then Cov(X , Y ) = 0 and Corr(X , Y ) = 0. In
fact, every function of X is uncorrelated with Y
⇒ If knowledge of X does not change the expected value of Y , then
X and Y must be uncorrelated
⇒ If X and Y are correlated, E (Y | X ) must depend on X
▶ But: If X and Y are uncorrelated, then E (Y | X ) could still depend
on X (Example: Y = X 2 ⇒ E (Y | X ) = X 2 )
⇒ The conditional expectation captures the nonlinear relationship
between X and Y that correlation analysis would miss entirely
Chen Tong (SOE&WISE) Risk Management September 12, 2024 40 / 62
Conditional Variance
▶ The variance of Y given X = x is given by
Var(Y | X = x) = E (Y − E (Y | X = x))2 | X = x
= E Y 2 | X = x − [E (Y | X = x)]2
▶ Example: Var( SAVING|INCOME ) = 400 + .25 INCOME
⇒ The variance in savings increases with income
("heteroscedasticity")
▶ If X and Y are independent, then Var(Y | X ) = Var(Y )
Chen Tong (SOE&WISE) Risk Management September 12, 2024 41 / 62
∗∗Transformation of R.V. (Random Variable)
▶ Y = r (X ) and X pdf
∼ fX
▶ the cdf of Y is
Z
G (y ) = Pr(Y ⩽ y ) = Pr(r (X ) ⩽ y ) = f (x)dx,
{x:r (X )⩽y }
▶ if r is monotonic, the pdf of Y is
ds(y )
g (y ) = f (s(y )) ,
dy
where s(·) is the inverse function of r . i.e. x = s(y ).
Chen Tong (SOE&WISE) Risk Management September 12, 2024 42 / 62
▶ For instance, compute the pdf of y if log(y ) ∼ N(µ, σ 2 )
Chen Tong (SOE&WISE) Risk Management September 12, 2024 43 / 62
Probability Theory
4. The Normal and Related Distributions
Chen Tong (SOE&WISE) Risk Management September 12, 2024 44 / 62
▶ The Normal Distribution
▶ The Standard Normal Distribution
▶ The Chi-Square Distribution
▶ The t-Distribution
▶ The F-Distribution
Chen Tong (SOE&WISE) Risk Management September 12, 2024 45 / 62
The Normal Distribution
▶ Assuming that random variables defined over populations are
normally distributed simplifies probability calculations
▶ The pdf of a normal random variable X is
1
√ exp −(x − µ)2 /2σ 2 ,
f (x) = −∞ < x < ∞
σ 2π
where µX = E (X ) and σX2 = Var(X )
▶ X is normally distributed with expected value µ and variance σ 2 ,
written as X ∼ N µ, σ 2
▶ Examples: Human heights, weights, test scores, county
unemployment rates, etc.
▶ Income has a log-normal distribution, i.e. Y = log(INCOME ) has a
normal distribution
Chen Tong (SOE&WISE) Risk Management September 12, 2024 46 / 62
Chen Tong (SOE&WISE) Risk Management September 12, 2024 47 / 62
The Moments of Normal Distribution
▶ For any non-negative integer p, the central moments is given by
(
0 if p is odd
E [(X − µ)p ] =
σ p (p − 1)!! if p is even
where n!!denotes the double factorial, that for even n, we have
n
2
Y
n!! = (2k) = n(n − 2)(n − 4) · · · 4 · 2,
k=1
while for odd n it is
n+1
2
Y
n!! = (2k − 1) = n(n − 2)(n − 4) · · · 3 · 1.
k=1
Chen Tong (SOE&WISE) Risk Management September 12, 2024 48 / 62
The (non)Central Moments of Normal Distribution
The non-central moments are defined by E [X p ], we have
Chen Tong (SOE&WISE) Risk Management September 12, 2024 49 / 62
The Standard Normal Distribution
▶ Special case of the normal distribution where the mean is zero
(µ = 0) and the variance is one σ 2 = σ = 1
▶ If a random variable Z has a standard normal distribution, then
Z ∼ N(0, 1)
▶ The pdf of a standard normal distribution is
1
ϕ(z) = √ exp −z 2 /2 ,
−∞ < z < ∞
2π
▶ The standard normal cumulative distribution function is denoted
Φ(z) = P(Z < z)
Chen Tong (SOE&WISE) Risk Management September 12, 2024 50 / 62
Chen Tong (SOE&WISE) Risk Management September 12, 2024 51 / 62
The Chi-Square Distribution
▶ The chi-square distribution is obtained from independent, standard
normal variables
▶ Let Zi (i = 1, 2, . . . , n) be independent random variables, each
distributed as standard normal. Then a new random variable may be
defined as the sum of the squares of Zi :
n
X
X = Zi2
i=1
Chen Tong (SOE&WISE) Risk Management September 12, 2024 52 / 62
The Chi-Square Distribution
▶ Then X has a chi-squared distribution with n degrees of freedom:
X ∼ χ2n
▶ The chi-square distribution is the "ideal" counterpart of the normal
distribution in situations where the random variable is non-negative
▶ The form of the distribution varies with the number of degrees of
freedom, i.e. the number of random variables Zi included in X
Chen Tong (SOE&WISE) Risk Management September 12, 2024 53 / 62
Chen Tong (SOE&WISE) Risk Management September 12, 2024 54 / 62
The t-Distribution
▶ The t-distribution can be obtained from a standard normal and a
chi-square random variable:
▶ Let Z have a standard normal distribution, let X have a chi-square
distribution with n degrees of freedom and assume that Z and X are
independent. Then the random variable
Z
T =p
X /n
has a t-distribution with ν degrees of freedom, T ∼ tn
▶ The shape of the t-distribution is similar to that of a normal
distribution, except that the t-distribution has more probability mass
in the tails
▶ As the degrees of freedom get large, the t-distribution approaches
the standard normal distribution
Chen Tong (SOE&WISE) Risk Management September 12, 2024 55 / 62
The pdf of t-Distribution
▶ The density function of t-distribution with n degrees of freedom is
−(n+1)/2
Γ n+1
2 x2
f (x) = √ 1+
nπΓ n2 n
where Γ(·) is the Gamma function.
▶ tn has heavier tails and the amount of probability mass in the tails is
controlled by the parameter n. For n = 1 the t distribution tn
becomes the standard Cauchy distribution, whereas for n → ∞ it
becomes the standard normal distribution N(0, 1).
Chen Tong (SOE&WISE) Risk Management September 12, 2024 56 / 62
Moments of the t-Distribution
▶ For n > 1, the raw moments of the t-distribution are
0 k odd
E Tk =
h
1 k+1
n−k k i
√πΓ n Γ 2 Γ 2 n 2 k even
(2)
where 0 < k < n.
▶ Moments of order n or higher do not exist.
Chen Tong (SOE&WISE) Risk Management September 12, 2024 57 / 62
Moments of the t-Distribution
▶ For a t-distribution with n degrees of freedom, the expected value is
n
0 if n > 1, and its variance is n−2 if n > 2. The skewness is 0 if
6
n > 3 and the excess kurtosis is n−4 if n > 4.
▶ The excess kurtosis is defined by kurtosis minus three.
Chen Tong (SOE&WISE) Risk Management September 12, 2024 58 / 62
Chen Tong (SOE&WISE) Risk Management September 12, 2024 59 / 62
The Standardized t-Distribution
▶ For n > 2, the Standardized t-Distribution is given by
T
ST = q , T ∼ tn
n
n−2
6
So we have its variance is 1, and the excess kurtosis is n−4 if n > 4.
▶ How to derive the pdf of standardized t-Distribution?
Chen Tong (SOE&WISE) Risk Management September 12, 2024 60 / 62
The F-Distribution
▶ Let X1 ∼ χ2k and X2 ∼ χ2k and assume that X1 and X2 are
1 2
independent. Then the random variable
(X1 /k1 )
F =
(X2 /k2 )
has an F-distribution with (k1 , k2 ) degrees of freedom, F ∼ Fk1 ,k2
▶ k1 - numerator degrees of freedom
▶ k2 - denominator degrees of freedom
Chen Tong (SOE&WISE) Risk Management September 12, 2024 61 / 62
Chen Tong (SOE&WISE) Risk Management September 12, 2024 62 / 62