Financial Engineering & Risk Management

Review of Basic Probability

M. Haugh G. Iyengar
Department of Industrial Engineering and Operations Research
Columbia University
Discrete Random Variables
Denition. The cumulative distribution function (CDF), F(), of a random
variable, X, is dened by
F(x) := P(X x).
Denition. A discrete random variable, X, has probability mass function (PMF),
p(), if p(x) 0 and for all events A we have
P(X A) =

xA
p(x).
Denition. The expected value of a discrete random variable, X, is given by
E[X] :=

i
x
i
p(x
i
).
Denition. The variance of any random variable, X, is dened as
Var(X) := E
_
(X E[X])
2

= E[X
2
] E[X]
2
.
2
The Binomial Distribution
We say X has a binomial distribution, or X Bin(n, p), if
P(X = r) =
_
n
r
_
p
r
(1 p)
nr
.
For example, X might represent the number of heads in n independent coin
tosses, where p = P(head). The mean and variance of the binomial distribution
satisfy
E[X] = np
Var(X) = np(1 p).
3
A Financial Application
Suppose a fund manager outperforms the market in a given year with
probability p and that she underperforms the market with probability 1 p.
She has a track record of 10 years and has outperformed the market in 8 of
the 10 years.
Moreover, performance in any one year is independent of performance in
other years.
Question: How likely is a track record as good as this if the fund manager had no
skill so that p = 1/2?
Answer: Let X be the number of outperforming years. Since the fund manager
has no skill, X Bin(n = 10, p = 1/2) and
P(X 8) =
n

r=8
_
n
r
_
p
r
(1 p)
nr
Question: Suppose there are M fund managers? How well should the best one do
over the 10-year period if none of them had any skill?
4
The Poisson Distribution
We say X has a Poisson() distribution if
P(X = r) =

r
e

r!
.
E[X] = and Var(X) = .
For example, the mean is calculated as
E[X] =

r=0
r P(X = r) =

r=0
r

r
e

r!
=

r=1
r

r
e

r!
=

r=1

r1
e

(r 1)!
=

r=0

r
e

r!
= .
5
Bayes Theorem
Let A and B be two events for which P(B) = 0. Then
P(A| B) =
P(A

B)
P(B)
=
P(B | A)P(A)
P(B)
=
P(B | A)P(A)

j
P(B | A
j
)P(A
j
)
where the A
j
s form a partition of the sample-space.
6
An Example: Tossing Two Fair 6-Sided Dice
6 7 8 9 10 11 12
5 6 7 8 9 10 11
4 5 6 7 8 9 10
Y
2
3 4 5 6 7 8 9
2 3 4 5 6 7 8
1 2 3 4 5 6 7
1 2 3 4 5 6
Y
1
Table : X = Y
1
+Y
2
Let Y
1
and Y
2
be the outcomes of tossing two fair dice independently of
one another.
Let X := Y
1
+ Y
2
. Question: What is P(Y
1
4 | X 8)?
7
Continuous Random Variables
Denition. A continuous random variable, X, has probability density function
(PDF), f (), if f (x) 0 and for all events A
P(X A) =
_
A
f (y) dy.
The CDF and PDF are related by
F(x) =
_
x

f (y) dy.
It is often convenient to observe that
P
_
X
_
x

2
, x +

2
__
f (x)
8
The Normal Distribution
We say X has a Normal distribution, or X N(,
2
), if
f (x) =
1

2
2
exp
_

(x )
2
2
2
_
.
The mean and variance of the normal distribution satisfy
E[X] =
Var(X) =
2
.
9
The Log-Normal Distribution
We say X has a log-normal distribution, or X LN(,
2
), if
log(X) N(,
2
).
The mean and variance of the log-normal distribution satisfy
E[X] = exp( +
2
/2)
Var(X) = exp(2 +
2
) (exp(
2
) 1).
The log-normal distribution plays a very important in nancial applications.
10
Financial Engineering & Risk Management
Review of Conditional Expectations and Variances
M. Haugh G. Iyengar
Department of Industrial Engineering and Operations Research
Columbia University
Conditional Expectations and Variances
Let X and Y be two random variables.
The conditional expectation identity says
E[X] = E[E[X|Y]]
and the conditional variance identity says
Var(X) = Var(E[X|Y]) + E[Var(X|Y)].
Note that E[X|Y] and Var(X|Y) are both functions of Y and are therefore
random variables themselves.
2
A Random Sum of Random Variables
Let W = X
1
+X
2
+. . . +X
N
where the X
i
s are IID with mean
x
and variance

2
x
, and where N is also a random variable, independent of the X
i
s.
Question: What is E[W]?
Answer: The conditional expectation identity implies
E[W] = E
_
E
_
N

i=1
X
i
| N
__
= E[N
x
] =
x
E[N] .
Question: What is Var(W)?
Answer: The conditional variance identity implies
Var(W) = Var(E[W|N]) + E[Var(W|N)]
= Var(
x
N) + E[N
2
x
]
=
2
x
Var(N) +
2
x
E[N].
3
An Example: Chickens and Eggs
A hen lays N eggs where N Poisson(). Each egg hatches and yields a chicken
with probability p, independently of the other eggs and N. Let K be the number
of chickens.
Question: What is E[K|N]?
Answer: We can use indicator functions to answer this question.
In particular, can write K =

N
i=1
1
H
i
where H
i
is the event that the i
th
egg
hatches. Therefore
1
H
i
=
_
1, if i
th
egg hatches;
0, otherwise.
Also clear that E[1
H
i
] = 1 p + 0 (1 p) = p so that
E[K|N] = E
_
N

i=1
1
H
i
| N
_
=
N

i=1
E[1
H
i
] = Np.
Conditional expectation formula then gives E[K] = E[E[K|N]] = E[Np] = p.
4
Financial Engineering & Risk Management
Review of Multivariate Distributions
M. Haugh G. Iyengar
Department of Industrial Engineering and Operations Research
Columbia University
Multivariate Distributions I
Let X = (X
1
. . . X
n
)

be an n-dimensional vector of random variables.

Denition. For all x = (x
1
, . . . , x
n
) R
n
, the joint cumulative distribution
function (CDF) of X satises
F
X
(x) = F
X
(x
1
, . . . , x
n
) = P(X
1
x
1
, . . . , X
n
x
n
).
Denition. For a xed i, the marginal CDF of X
i
satises
F
X
i
(x
i
) = F
X
(, . . . , , x
i
, , . . . ).
It is straightforward to generalize the previous denition to joint marginal
distributions. For example, the joint marginal distribution of X
i
and X
j
satises
F
ij
(x
i
, x
j
) = F
X
(, . . . , , x
i
, , . . . , , x
j
, , . . . ).
We also say that X has joint PDF f
X
(, . . . , ) if
F
X
(x
1
, . . . , x
n
) =
_
x
1

_
x
n

f
X
(u
1
, . . . , u
n
) du
1
. . . du
n
.
2
Multivariate Distributions II
Denition. If X
1
= (X
1
, . . . X
k
)

and X
2
= (X
k+1
. . . X
n
)

is a partition of
X then the conditional CDF of X
2
given X
1
satises
F
X
2
|X
1
(x
2
| x
1
) = P(X
2
x
2
| X
1
= x
1
).
If X has a PDF, f
X
(), then the conditional PDF of X
2
given X
1
satises
f
X
2
|X
1
(x
2
| x
1
) =
f
X
(x)
f
X
1
(x
1
)
=
f
X
1
|X
2
(x
1
| x
2
)f
X
2
(x
2
)
f
X
1
(x
1
)
(1)
and the conditional CDF is then given by
F
X
2
|X
1
(x
2
|x
1
) =
_
x
k+1

_
x
n

f
X
(x
1
, . . . , x
k
, u
k+1
, . . . , u
n
)
f
X
1
(x
1
)
du
k+1
. . . du
n
where f
X
1
() is the joint marginal PDF of X
1
which is given by
f
X
1
(x
1
, . . . , x
k
) =
_

f
X
(x
1
, . . . , x
k
, u
k+1
, . . . , u
n
) du
k+1
. . . du
n
.
3
Independence
Denition. We say the collection X is independent if the joint CDF can be
factored into the product of the marginal CDFs so that
F
X
(x
1
, . . . , x
n
) = F
X
1
(x
1
) . . . F
X
n
(x
n
).
If X has a PDF, f
X
() then independence implies that the PDF also factorizes
into the product of marginal PDFs so that
f
X
(x) = f
X
1
(x
1
) . . . f
X
n
(x
n
).
Can also see from (1) that if X
1
and X
2
are independent then
f
X
2
|X
1
(x
2
| x
1
) =
f
X
(x)
f
X
1
(x
1
)
=
f
X
1
(x
1
)f
X
2
(x
2
)
f
X
1
(x
1
)
= f
X
2
(x
2
)
so having information about X
1
tells you nothing about X
2
.
4
Implications of Independence
Let X and Y be independent random variables. Then for any events, A and B,
P(X A, Y B) = P(X A) P(Y B) (2)
More generally, for any function, f () and g(), independence of X and Y implies
E[f (X)g(Y)] = E[f (X)]E[g(Y)]. (3)
In fact, (2) follows from (3) since
P(X A, Y B) = E
_
1
{XA}
1
{YB}

= E
_
1
{XA}

E
_
1
{YB}

by (3)
= P(X A) P(Y B) .
5
Implications of Independence
More generally, if X
1
, . . . X
n
are independent random variables then
E[f
1
(X
1
)f
2
(X
2
) f
n
(X
n
)] = E[f
1
(X
1
)]E[f
2
(X
2
)] E[f
n
(X
n
)].
Random variables can also be conditionally independent. For example, we say X
and Y are conditionally independent given Z if
E[f (X)g(Y) | Z] = E[f (X) | Z] E[g(Y) | Z].
used in the (in)famous Gaussian copula model for pricing CDOs!
In particular, let D
i
be the event that the i
th
bond in a portfolio defaults.
Not reasonable to assume that the D
i
s are independent. Why?
But maybe they are conditionally independent given Z so that
P(D
1
, . . . , D
n
| Z) = P(D
1
| Z) P(D
n
| Z)
often easy to compute this.
6
The Mean Vector and Covariance Matrix
The mean vector of X is given by
E[X] := (E[X
1
] . . . E[X
n
])

and the covariance matrix of X satises

:= Cov(X) := E
_
(XE[X]) (XE[X])

so that the (i, j)

th
element of is simply the covariance of X
i
and X
j
.
The covariance matrix is symmetric and its diagonal elements satisfy
i,i
0.
It is also positive semi-denite so that x

x 0 for all x R
n
.
The correlation matrix, (X), has (i, j)
th
element
ij
:= Corr(X
i
, X
j
)
- it is also symmetric, positive semi-denite and has 1s along the diagonal.
7
Variances and Covariances
For any matrix A R
kn
and vector a R
k
we have
E[AX+a] = AE[X] + a (4)
Cov(AX+a) = A Cov(X) A

. (5)
Note that (5) implies
Var(aX + bY) = a
2
Var(X) + b
2
Var(Y) + 2ab Cov(X, Y).
If X and Y independent, then Cov(X, Y) = 0
but converse not true in general.
8
Financial Engineering & Risk Management
The Multivariate Normal Distribution
M. Haugh G. Iyengar
Department of Industrial Engineering and Operations Research
Columbia University
The Multivariate Normal Distribution I
If the n-dimensional vector X is multivariate normal with mean vector and
covariance matrix then we write
X MN
n
(, ).
The PDF of X is given by
f
X
(x) =
1
(2)
n/2
||
1/2
e

1
2
(x)

1
(x)
where | | denotes the determinant.
Standard multivariate normal has = 0 and = I
n
, the n n identity matrix
- in this case the X
i
s are independent.
The moment generating function (MGF) of X satises

X
(s) = E
_
e
s

X
_
= e
s

+
1
2
s

s
.
2
The Multivariate Normal Distribution II
Recall our partition of X into X
1
= (X
1
. . . X
k
)

and X
2
= (X
k+1
. . . X
n
)

.
Can extend this notation naturally so that
=
_

1

2
_
and =
_

11

12

21

22
_
.
are the mean vector and covariance matrix of (X
1
, X
2
).
Then have following results on marginal and conditional distributions of X:
Marginal Distribution
The marginal distribution of a multivariate normal random vector is itself normal.
In particular, X
i
MN(
i
,
ii
), for i = 1, 2.
3
The Bivariate Normal PDF
The Bivariate Normal PDF
4
The Multivariate Normal Distribution III
Conditional Distribution
Assuming is positive denite, the conditional distribution of a multivariate
normal distribution is also a multivariate normal distribution. In particular,
X
2
| X
1
= x
1
MN(
2.1
,
2.1
)
where
2.1
=
2
+
21

1
11
(x
1

1
) and
2.1
=
22

21

1
11

12
.
Linear Combinations
A linear combination, AX+a, of a multivariate normal random vector, X, is
normally distributed with mean vector, AE[X] +a, and covariance matrix,
A Cov(X) A

.
5
Financial Engineering & Risk Management
Introduction to Martingales
M. Haugh G. Iyengar
Department of Industrial Engineering and Operations Research
Columbia University
Martingales
Denition. A random process, {X
n
: 0 n }, is a martingale with respect
to the information ltration, F
n
, and probability distribution, P, if
1. E
P
[|X
n
|] < for all n 0
2. E
P
[X
n+m
|F
n
] = X
n
for all n, m 0.
Martingales are used to model fair games and have a rich history in the modeling
of gambling problems.
We dene a submartingale by replacing condition #2 with
E
P
[X
n+m
|F
n
] X
n
for all n, m 0.
And we dene a supermartingale by replacing condition #2 with
E
P
[X
n+m
|F
n
] X
n
for all n, m 0.
A martingale is both a submartingale and a supermartingale.
2
Constructing a Martingale from a Random Walk
Let S
n
:=

n
i=1
X
i
be a random walk where the X
i
s are IID with mean .
Let M
n
:= S
n
n. Then M
n
is a martingale because:
E
n
[M
n+m
] = E
n
_
n+m

i=1
X
i
(n + m)
_
= E
n
_
n+m

i=1
X
i
_
(n + m)
=
n

i=1
X
i
+ E
n
_
n+m

i=n+1
X
i
_
(n + m)
=
n

i=1
X
i
+ m (n + m) = M
n
.
3
A Martingale Betting Strategy
Let X
1
, X
2
, . . . be IID random variables with
P(X
i
= 1) = P(X
i
= 1) =
1
2
.
Can imagine X
i
representing the result of coin-ipping game:
Win $1 if coin comes up heads
Lose $1 if coin comes up tails
Consider now a doubling strategy where we keep doubling the bet until we
eventually win. Once we win, we stop and our initial bet is $1.
First note that size of bet on n
th
play is 2
n1
assuming were still playing at time n.
Let W
n
denote total winnings after n coin tosses assuming W
0
= 0.
Then W
n
is a martingale!
4
A Martingale Betting Strategy
To see this, rst note that W
n
{1, 2
n
+ 1} for all n. Why?
1. Suppose we win for rst time on n
th
bet. Then
W
n
=
_
1 + 2 + + 2
n2
_
+ 2
n1
=
_
2
n1
1
_
+ 2
n1
= 1
2. If we have not yet won after n bets then
W
n
=
_
1 + 2 + + 2
n1
_
= 2
n
+ 1.
To show W
n
is a martingale only need to show E[W
n+1
| W
n
] = W
n
then follows by iterated expectations that E[W
n+m
| W
n
] = W
n
.
5
A Martingale Betting Strategy
There are two cases to consider:
1: W
n
= 1: then P(W
n+1
= 1|W
n
= 1) = 1 so
E[W
n+1
| W
n
= 1] = 1 = W
n
(6)
2: W
n
= 2
n
+ 1: bet 2
n
on (n + 1)
th
toss so W
n+1
{1, 2
n+1
+ 1}.
Clear that
P(W
n+1
= 1 | W
n
= 2
n
+ 1) = 1/2
P(W
n+1
= 2
n+1
+ 1 | W
n
= 2
n
+ 1) = 1/2
so that
E[W
n+1
| W
n
= 2
n
+ 1] = (1/2)1 + (1/2)(2
n+1
+ 1)
= 2
n
+ 1 = W
n
. (7)
From (6) and (7) we see that E[W
n+1
| W
n
] = W
n
.
6
Polyas Urn
Consider an urn which contains red balls and green balls.
Initially there is just one green ball and one red ball in the urn.
At each time step a ball is chosen randomly from the urn:
1. If ball is red, then its returned to the urn with an additional red ball.
2. If ball is green, then its returned to the urn with an additional green ball.
Let X
n
denote the number of red balls in the urn after n draws. Then
P(X
n+1
= k + 1 | X
n
= k) =
k
n + 2
P(X
n+1
= k | X
n
= k) =
n + 2 k
n + 2
.
Show that M
n
:= X
n
/(n + 2) is a martingale.
(These martingale examples taken from Introduction to Stochastic Processes
(Chapman & Hall) by Gregory F. Lawler.)
7
Financial Engineering & Risk Management
Introduction to Brownian Motion
M. Haugh G. Iyengar
Department of Industrial Engineering and Operations Research
Columbia University
Brownian Motion
Denition. We say that a random process, {X
t
: t 0}, is a Brownian motion
with parameters (, ) if
1. For 0 < t
1
< t
2
< . . . < t
n1
< t
n
(X
t
2
X
t
1
), (X
t
3
X
t
2
), . . . , (X
t
n
X
t
n1
)
are mutually independent.
2. For s > 0, X
t+s
X
t
N(s,
2
s) and
3. X
t
is a continuous function of t.
We say that X
t
is a B(, ) Brownian motion with drift and volatility
Property #1 is often called the independent increments property.
Remark. Bachelier (1900) and Einstein (1905) were the rst to explore Brownian
motion from a mathematical viewpoint whereas Wiener (1920s) was the rst to
show that it actually exists as a well-dened mathematical entity.
2
Standard Brownian Motion
When = 0 and = 1 we have a standard Brownian motion (SBM).
We will use W
t
to denote a SBM and we always assume that W
0
= 0.
Note that if X
t
B(, ) and X
0
= x then we can write
X
t
= x +t +W
t
(8)
where W
t
is an SBM. Therefore see that X
t
N(x +t,
2
t).
3
Sample Paths of Brownian Motion
4
Information Filtrations
For any random process we will use F
t
to denote the information available
at time t
- the set {F
t
}
t0
is then the information ltration
- so E[ | F
t
] denotes an expectation conditional on time t information available.
Will usually write E[ | F
t
] as E
t
[ ].
Important Fact: The independent increments property of Brownian motion
implies that any function of W
t+s
W
t
is independent of F
t
and that
(W
t+s
W
t
) N(0, s).
5
A Brownian Motion Calculation
Question: What is E
0
[W
t+s
W
s
]?
Answer: We can use a version of the conditional expectation identity to obtain
E
0
[W
t+s
W
s
] = E
0
[(W
t+s
W
s
+ W
s
) W
s
]
= E
0
[(W
t+s
W
s
) W
s
] + E
0
_
W
2
s

. (9)
Now we know (why?) E
0
[W
2
s
] = s.
To calculate rst term on r.h.s. of (9) a version of the conditional expectation
identity implies
E
0
[(W
t+s
W
s
) W
s
] = E
0
[E
s
[(W
t+s
W
s
) W
s
]]
= E
0
[W
s
E
s
[(W
t+s
W
s
)]]
= E
0
[W
s
0]
= 0.
Therefore obtain E
0
[W
t+s
W
s
] = s.
6
Financial Engineering & Risk Management
Geometric Brownian Motion
M. Haugh G. Iyengar
Department of Industrial Engineering and Operations Research
Columbia University
Geometric Brownian Motion
Denition. We say that a random process, X
t
, is a geometric Brownian motion
(GBM) if for all t 0
X
t
= e
_

2
2
_
t + W
t
where W
t
is a standard Brownian motion.
We call the drift, the volatility and write X
t
GBM(, ).
Note that
X
t+s
= X
0
e
_

2
2
_
(t+s) + W
t+s
= X
0
e
_

2
2
_
t + W
t
+
_

2
2
_
s + (W
t+s
W
t
)
= X
t
e
_

2
2
_
s + (W
t+s
W
t
)
(10)
a representation that is very useful for simulating security prices.
2
Geometric Brownian Motion
Question: Suppose X
t
GBM(, ). What is E
t
[X
t+s
]?
Answer: From (10) we have
E
t
[X
t+s
] = E
t
_
X
t
e
_

2
2
_
s + (W
t+s
W
t
)
_
= X
t
e
_

2
2
_
s
E
t
_
e
(W
t+s
W
t
)
_
= X
t
e
_

2
2
_
s
e

2
2
s
= e
s
X
t
so the expected growth rate of X
t
is .
3
Sample Paths of Geometric Brownian Motion
4
Geometric Brownian Motion
The following properties of GBM follow immediately from the denition of BM:
1. Fix t
1
, t
2
, . . . , t
n
. Then
X
t
2
X
t
1
,
X
t
3
X
t
2
, . . . ,
X
t
n
X
t
n1
are mutually independent.
2. Paths of X
t
are continuous as a function of t, i.e., they do not jump.
3. For s > 0, log
_
X
t+s
X
t
_
N
_
(

2
2
)s,
2
s
_
.
5
Modeling Stock Prices as GBM
Suppose X
t
GBM(, ). Then clear that:
1. If X
t
> 0, then X
t+s
is always positive for any s > 0.
- so limited liability of stock price is not violated.
2. The distribution of X
t+s
/X
t
only depends on s and not on X
t
These properties suggest that GBM might be a reasonable model for stock prices.
Indeed it is the underlying model for the famous Black-Scholes option formula.
6