Stochastic Calculus for Finance
Module 02: Stochastic Integration and Its
Connection to Finance
Dr. Chandan Pal
Department of Mathematics
Indian Institute of Technology Guwahati
Stochastic Integration
Definition
Let (Ω, F, P) be a probability space. Let W (·) be a Brownian motion
defined on it. Let Ft be a Brownian filtration. For T > 0, define the
space L2ad ([0, T ] × Ω) to denote the space of all stochastic processes
f (t, ω), 0 ≤ t ≤ T , ω ∈ Ω satisfying the following two properties:
f (t, ω) is adapted to the filtration {Ft }.
RT
0
E|f (t)|2 dt < ∞.
Simple Stochastic Process
Definition
Let (Ω, F, P) be a probability space. Let W (·) be a Brownian motion
defined on it. Let Ft be a Brownian filtration. A stochastic process
f (t), t ≥ 0 is said to be a simple stochastic process if there is a finite
sequence of numbers 0 = t0 , t1 , . . . , tn = T and square integrable
random variables ξ0 , ξ1 , . . . , ξn−1 (E[ξi2 ] < ∞) such that
n
X
f (t, ω) = ξi−1 1[ti−1 ,ti ) (t) ,
i=1
where {t0 , t1 , . . . , tn } is a partition of [0, T ] and ξi−1 is Fti−1
2
measurable and E(ξi−1 ) < ∞.
Stochastic Integration
If f is a simple stochastic process in L2ad ([0, T ] × Ω), i.e.,
n
X
f (t, ω) = ξi−1 1[ti−1 ,ti ) (t) ,
i=1
where {t0 , t1 , . . . , tn } is a partition of [0, T ] and ξi−1 is Fti−1
2
measurable and E(ξi−1 ) < ∞.
The Ito integration of f on [0, t], tk−1 ≤ t ≤ tk is given by
Z t k−1
X
It (f ) = f (t)dWt = ξi−1 (W (ti ) − W (ti−1 ))
0 i=1
+ ξk−1 (W (t) − W (tk−1 )) .
Stochastic integral as a gain process
Let W (t) be the price per share of an asset at time t
Think of t0 , t1 , . . . , tn−1 as the trading dates
f (t0 ), f (t1 ), . . . , f (tn−1 ) as the position (number of shares) taken
in the asset at each trading date and held to the next trading
date
The gain It (f ) from trading at each time t is given by
It (f ) = f (t0 )(W (t) − W (t0 )) = f (0)W (t) if t0 < t ≤ t1
It (f ) = f (0)W (t1 ) + f (t1 )(W (t) − W (t1 )) if t1 < t ≤ t2
It (f ) = f (0)W (t1 ) + f (t1 )(W (t2 ) − W (t1 )) + f (t2 )(W (t) − W (t2 ))
In general, if tk < t ≤ tk+1 then
Pk−1
IRt (f ) = i=0 f (ti )(W (ti+1 ) − W (ti )) + f (tk )(W (t) − W (tk )) :=
t
0
f (s)dW (s)
Ito process
Definition
Let W (t), t ≥ 0 be a Brownian motion and let Ft , t ≥ 0 be a
filtration for the Brownian motion. An Ito process is a stochastic
process of the form
Z t Z t
X (t) = X (0) + ∆(u)dW (u) + Θ(u)du ,
0 0
where
Rt ∆(u) and Θ(u) are
R t adapted stochastic processes such that
2
0
E∆ (u)du < ∞ and 0 |Θ(u)|du < ∞ for all t ≥ 0.
Lemma
The quadratic variation of the above Ito process is
Z t
[X , X ](t) = ∆2 (u)du .
0
Ito process
Remark: Symbolically, dX (t) = ∆(t)dW (t) + Θ(t)dt, so
dXt dXt = ∆2 (t)dWt dWt +Θ2 (t)dtdt +2∆(t)Θ(t)dWt dt = ∆2 (t)dt .
Definition
Let X (t), t ≥ 0 be an Ito process.
R t Let Γ(t), t ≥ 0 be an adapted
process. Further suppose that 0 E(Γ2 (u)∆2 (u))du and
Rt
0
|Γ(u)Θ(u)|du are finite for all t ≥ 0. We define the integral with
respect to an Ito process by,
Z t Z t Z t
Γ(u)dX (u) = Γ(u)∆(u)dW (u) + Γ(u)Θ(u)du .
0 0 0
Ito’s Formula
Theorem:(Ito-Doeblin Formula for Ito process) Let X (t), t ≥ 0 be
an Ito process. Let f (t, x) be a function such that the partial
derivatives ft (t, x), fx (t, x) and fxx (t, x) all exist and are continuous.
Then for every T ≥ 0,
f (T , X (T )) − f (0, X (0))
Z T Z T
= ft (t, X (t))dt + fx (t, X (t))dX (t)
0 0
1 T
Z
+ fxx (t, X (t))dX (t)dX (t)
2 0
Z T Z T
= ft (t, X (t))dt + fx (t, X (t))∆(t)dW (t)
0 0
Z T
1 T
Z
+ fx (t, X (t))Θ(t)dt + fxx (t, X (t))∆2 (t)dt .
0 2 0
Exponential Martingale:
Theorem
Let W (t), t ≥ 0 be a Brownian motion with filtration Ft , t ≥ 0 and
let σ be a constant. Then the process
1 2
Z (t) = e σW (t)− 2 σ t ,
is a martingale with respect to Ft .
Generalized geometric Brownian motion
Let W (t), t ≥ 0 be a Brownian motion and let Ft , t ≥ 0 be a
filtration for the Brownian motion and let α(u) and σ(u) be
adapted stochastic processes.
Define the Ito process
Z t Z t
1
X (t) = X (0) + σ(u)dW (u) + (α(u) − σ 2 (u))du .
0 0 2
Then
1
dX (t) = σ(t)dW (t) + (α(t) − σ 2 (t))dt ,
2
and
dX (t)dX (t) = σ 2 (t)dW (t)dW (t) = σ 2 (t)dt .
Consider an asset price process given by
nZ t Z t
1 o
S(t) = S(0) exp σ(u)dW (u) + (α(u) − σ 2 (u))du ,
0 0 2
where S(0) is nonrandom and positive.
Generalized geometric Brownian motion
Problem 1:
nZ t Z t
1 2
o
S(t) = S(0) exp σ(u)dW (u) − σ (u)du
0 2 0
is a martingale with respect to Brownian filtration.
Problem 2: Let W (·) be a Brownian
Rt motion and let σ(t) be a non
random function. Define I (t) = 0 σ(u)dW (u). Then show that I (t)
Rt
is a normal random variable with mean 0 and variance 0 σ 2 (u)du.