Lecture 6: Wiener Processes

and Its Lemma

Based on Chapter 13 from

John C. Hull (2012). Options, Futures &
Other Derivatives, 8th ed., Prentice Hall.
Stochastic Processes
Any variable whose value changes over time in an uncertain
way is said to follow a stochastic process.
Discrete time process: variable value can change only at certain xed
points in time.
Continuous-time process: variable value can change at any time.
Continuous-variable process: the underlying variable can take any
value (within a certain range).
Discrete-variable process: the underlying variable can only take
certain discrete values.
This chapter develops a continuous-variable, continuous-time
stochastic process for stock prices.
First step for understanding the pricing of options/derivatives.

Financial Mathematics 2
Examples
1. Each day a stock price
increases by $1 with probability 30%
stays the same with probability 50%
decreases by $1 with probability 20%

2. Each day a stock price change is drawn from a

normal distribution with mean $0.2 and standard
deviation $1.

Financial Mathematics 3
Markov Processes
In a Markov process, future movements in a variable depend
only on where we are now, not on the history of how we got to
where we are now.
Is the process followed by the temperature at a certain place Markov?
We assume that stock prices follow Markov processes.
This is consistent with the weak-form market efficiency:
It is impossible to produce consistently superior returns with a
trading rule based on the past history of stock prices.
In other words, technical analysis does not work.
There is very little evidence that interpreting charts of the past
history of stock prices could make above-average returns.
Financial Mathematics 4
Example 3
A stocks price (St) follows a Markov process, which is
(assumed to be) stationary, i.e., the parameters of the process do
not change over time.
Currently the stock price is 40: S0 = 40. At the end of 1 year the
stock price (S1) will have a normal probability distribution with
mean 40 and standard deviation 10: S1 ~ (40, 102) = (40, 100).
Questions: what is the probability distribution of the stock price
at the end of:
A) 2 years? B) years? C) years? D) t years?
(40, 200) (40, 50) (40, 25) (40, 100 t)
Taking limits we have defined a continuous stochastic process.
Financial Mathematics 5
Variances & Standard Deviations
In Markov processes, changes in successive periods
of time are independent.
This means that variances are additive.
But standard deviations are not additive.
In the above Example 3 it is correct to say that the
variance is 100 per year; 200 in 2 years; 50 in
years; 25 in years; 100t in t years.
Strictly speaking, it is not correct to say that the standard
deviation is 10 per year.
The standard deviation in years is 5, not 10/4 = 2.5!
Financial Mathematics 6
Wiener Processes (special Markov processes)
Define (, 2) as a normal distribution with mean
and variance 2. [ = standard deviation]
A variable z = z(t) follows a Wiener process if:
The change in z in a small interval of time t is z.
z = t , where follows (0, 1): ~(0, 1).
z follows (0, t): z ~(0, t).
The values of z for any two different (non-overlapping)
periods of time are independent.

Financial Mathematics 7
 Properties of a Wiener Process
Mean of z is 0.
Variance of z is t.
Standard deviation of z is t.
6A) Change in z from time 0 to time T, z(T) z(0), can be
regarded as the sum of the small changes (zi) in z in N
small time intervals of length t, where N = T/t.
Mean of [z(T) z(0)] is 0.
Variance of [z(T) z(0)] is T [= Nt].
Standard deviation of [z(T) z(0)] is T.
Financial Mathematics 8
Example 4
Suppose that the value, z, of a variable that follows a Wiener
process is initially 25 and that time is measured in years.
At the end of 3-month = years, the value of the variable is normally
distributed with a mean of 25 and a standard deviation of 0.5 = 1 / 4 .
At the end of 1 year, the value of the variable is normally distributed with a
mean of 25 and a standard deviation of 1.0.
At the end of 5 years, it is normally distributed with a mean of 25 and a
standard deviation of 5 , or 2.236.
Our uncertainty about the value of the variable at a certain time
in the future, as measured by its standard deviation, increases as
the square root ( ) of how far we are looking ahead.

Financial Mathematics 9
Generalized Wiener Processes
A Wiener process has a drift rate (i.e., average
change per unit time) of 0 and a variance rate of 1.
In a generalized Wiener process x = x(t), the drift
rate and the variance rate can be set equal to any
chosen constants:
x = a t + b z a t + b t
Mean change in x per unit time (drift rate) is a.
Variance of change in x per unit time is b2.

Financial Mathematics 10
Taking Limits . . .
What does an expression involving dz and dt mean?
It should be interpreted as meaning that the
corresponding expression involving z and t is true
in the limit as t tends to zero.
As t 0, y = a t tends to dy = a dt.
As t 0, x = a t + b z dx = a dt + b dz.
In this respect, stochastic calculus is analogous to
ordinary calculus.
Financial Mathematics 11
Understanding the Generalized Wiener Processes
Consider the two components in dx = a dt + b dz separately.
The a dt term implies that x has an expected drift rate of a per
unit of time. Without the b dz term, the equation is dx = a dt,
implying that x = x0 + at, where x0 is the value of x at time 0. In
a period of time of length T, x increases by an amount aT.
The b dz term can be regarded as adding noise or variability to
the (linear trending) path followed by x. The amount of this
noise or variability is b times a Wiener process. A Wiener
process has a variance rate per unit time of 1.0. It follows that b
times a Wiener process has a variance rate per unit time of b2.
In a small time interval t, the change x in the value of x is
given by x = a t + b z, where z = t and ~(0, 1).
Financial Mathematics 12
Example 3 Revisited
A stock price (S) starts at 40 and has a probability
distribution of (40,100) at the end of the year.
If we assume the stochastic process is Markov with
no drift (a = 0), then the process is Wiener:
dS = 10dz
If the stock price is expected to grow by $8 on
average during the year (a = 8), so that the year-end
distribution is (48,100), the process would be
dS = 8 dt + 10 dz
Financial Mathematics 13
It Process
In an It process x = x(t), the drift rate and the variance
rate are functions of time (t):
dx = a(x, t) dt + b(x, t) dz
This is still Markov because the change in x at time t depends
only on the value of x at time t, not on its history.
The discrete time equivalent :
x = a(x, t)t + b(x, t) t
is true (for changes from t to t + t).
In the limit as t tends to zero, we have the continuous model.

Financial Mathematics 14
Generalized Wiener Process Not Appropriate for Stocks
Does a stock price S = S(t) follow a generalized Wiener process
(with constant expected drift rate and constant variance rate)?
No! This model fails to capture a key aspect of stock prices, i.e.,
the expected percentage return (S/S) [not the expected drift
rate (S/t)] required by investors from a stock is independent
of the stocks price.
If investors require a 14% annual expected return when the stock price is
$10, they will also require a 14% annual expected return when it is $50.
So the constant expected drift rate assumption is inappropriate and needs
to be replaced by that the expected return (i.e., expected drift divided by
the stock price) is constant: S/S = , where the constant parameter is
the expected rate of return on the stock, expressed in decimal form.
Financial Mathematics 15
Generalized Wiener Process Not Appropriate for Stocks
In practice, of course, there is also uncertainty. A reasonable
assumption is that the variability of the percentage return [not
the drift (rate)] in a short period of time, t, is the same
regardless of the stock price.
An investor is just as uncertain of the percentage return when the stock
price is $50 as when it is $10. Or the standard deviation of price change
in a short period of time t should be proportional to the stock price.
As a summary, we have for a stock price that:
Its expected percentage change in a short period of time remains
constant (not its expected absolute change or drift rate).
Our uncertainty as to the size of future stock price movements is
proportional to the level of the stock price.
Financial Mathematics 16
An It Process for Stock Prices
The above arguments lead to the most widely used
model of stock price behavior:
6B) dS/S = dt + dz or dS = S dt + S dz
where is the expected return and is the volatility.
The process is also known as geometric Brownian motion.
The discrete time equivalent is
S/S = t + t or S = S t + S t
where ~(0, 1). So S/S ~ ( t, 2t).
Financial Mathematics 17
Monte Carlo Simulation
We can sample random paths for the stock price by sampling
values for ~ (0, 1).
Suppose = 0.15, = 0.30, and t = 1 week (= 1/52 or 0.0192
years), then S = 0.150.0192S + 0.30S 0.0192 .
Or S = 0.00288S + 0.0416S.
Monte Carlo simulation: sampling one path
Week Starting price S0 Random sample for Price change S
1 100.00 0.52 2.45
2 102.45 1.44 6.43
3 108.88 -0.86 -3.58
4 105.30 1.46 6.70
5 112.00 -0.69 -2.89
6 109.11 -0.74 -3.04
Financial Mathematics 18
Correlated Processes
Suppose dz1 and dz2 are Wiener processes with
correlation .
Then z1 = 1 t and z2 = 2 t ,
where 1 and 2 are random samples from a
bivariate standard normal distribution where the
correlation is .
Then dx1 = a1 dt + b1 dz1 and dx2 = a2 dt + b2 dz2
Or x1 = a1t + b11 t and x2 = a2t + b22 t .

Financial Mathematics 19
Its Lemma
If we know the stochastic process followed by x, Its lemma
tells us the stochastic process followed by a function G(x, t).
When x is It: dx = a(x, t) dt + b(x, t) dz, then:
G G 2G 2 G
6C) dG ( a b ) dt bdz
x t x 2
x
or dG = [( G/ x)a + G/ t + 0.5( 2G/ x2)b2]dt + ( G/ x)bdz.
Since a derivative is a function of the price of the underlying
asset and time, Its lemma plays an important part in the
analysis of derivatives.
y 2
y
Note: y/ x = and y/ x = 2 , which are just for convenience.
2 2
x x

Financial Mathematics 20
 Indication of Why Its Lemma Is True
A Taylors series expansion of G = G(x) gives
G 2G 2
G x x
x x 2

or G = ( G/ x)x + 0.5( 2G/ x2)x2 +

A Taylors series expansion of G = G(x, t) gives
G G 2G 2 2G 2G 2
G x t x xt t
x t x 2
xt t 2

or G = ( G/ x)x + ( G/ t)t
+ 0.5( 2G/ x2)x2 + [ 2G/( x t)]xt + 0.5( 2G/ t2)t2 +

Financial Mathematics 21
Ignoring Terms of Higher Order Than t
In ordinary calculus we have
G G
G x t
x t
or G = ( G/ x)x + ( G/ t)t.
In stochastic calculus this becomes
G G 2G
G x t x 2

x t x2

or G = ( G/ x)x + ( G/ t)t + 0.5( 2G/ x2)x2,

because x has a component which is of order t .
Financial Mathematics 22
 Substituting for x
Suppose dx = a(x, t) dt + b(x, t) dz,
so that x = a t + b t .
Then ignoring terms of higher order than t,
G = ( G/ x)x + ( G/ t)t + 0.5( 2G/ x2)b22t.
Or
G G 2G 2 2
G x t b t
x t x 2

Financial Mathematics 23
 The 2t Term
Since ~ (0, 1), E() = 0,
Var() = E(2) [E()]2 = 1,
E(2) = 1.
It follows that E(2t) = t E(2) = t.
The variance of 2t is proportional to t2 and can be
ignored. Hence
G G 2G 2
G x t b t
x t x 2

or G = ( G/ x)x + ( G/ t)t + 0.5( 2G/ x2)b2t.

Financial Mathematics 24
 Taking Limits
Taking limits: G G 2G 2
dG dx dt b dt
x t x 2

or dG = ( G/ x)dx + ( G/ t)dt + 0.5( 2G/ x2)b2dt.

Substituting: dx = a(x, t) dt + b(x, t) dz.
Finally we get:
G G 2G 2 G
6C) dG ( a b ) dt bdz
x t x 2
x
or dG = [( G/ x)a + G/ t + 0.5( 2G/ x2)b2]dt + ( G/ x)bdz.
This is just Its Lemma [equation 6C)].

Financial Mathematics 25
Application of Its Lemma
to a Stock Price Process
The stock price process {S = St} is geometric Brownian motion:
6B) dS = S dt + S dz. [a = S and b = S ]
For a function G of S and t, G = G(S, t):
G G 2G 2 2 G
6D) dG ( S S t S 2 S )dt S Sdz
or dG = [(G/S)S + G/t + 0.5( 2G/S2)2S2]dt + (G/S)Sdz.

Financial Mathematics 26
 Application to Forward Contracts
The forward price F of a stock for a contract maturing at time T:
Price at time 0: F0 = S0erT. Price at time t: F = Ser(T-t).
F/ S = er(T-t), 2F/ S2 = 0; F/ t = -rSer(T-t).
From Its Lemma [equation 6D)]:
dF = [( F/S)S + F/t + 0.5( 2F/S2)2S2]dt + (F/S)Sdz
= [er(T-t) S rSer(T-t) + 0.5(0)2S2]dt + er(T-t) S dz
= ( - r)Ser(T-t) dt + Ser(T-t)dz = ( - r)Fdt + Fdz.
Like S, the forward price F follows geometric Brownian motion. It
has an expected growth rate of - r rather than : the growth rate
in F is the excess return of S over the risk-free rate (r).
Financial Mathematics 27
 The Lognormal Process
Consider the log of a stock price process (S): G = ln(S).
G/ S = 1/S, 2G/ S2 = -1/S2; G/ t = 0.
dG = [( G/S)S + G/t + 0.5( 2G/S2)2S2]dt + (G/S)Sdz
= [(1/S)S + 0 + 0.5(-1/S2)2S2]dt + (1/S) S dz
= ( - 0.52) dt + dz.
This shows that ln(S) follows a general Wiener process
with constant drift rate - 0.52 and constant variance 2.
So the change in ln(S) between time 0 and time T is normally
distributed, with mean ( - 0.52)T and variance 2T:
ln(ST) ln(S0) ~ [( - 0.52)T, 2T]
or ln(ST) ~ [ln(S0) + ( - 0.52)T, 2T]
Financial Mathematics 28