0% found this document useful (0 votes)
68 views25 pages

It o Calculus: Types of Derivatives

1) Itô calculus provides stochastic counterparts to derivatives and integrals for functions of stochastic processes. 2) Itô's lemma gives the stochastic version of the chain rule for functions of stochastic processes like stock prices that follow stochastic differential equations (SDEs). 3) Applying Itô's lemma, the SDE for a derivative can be derived once the SDE of the underlying asset is known. Itô's formula is thus useful for finding SDEs and evaluating stochastic integrals.

Uploaded by

Norbert Durand
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
68 views25 pages

It o Calculus: Types of Derivatives

1) Itô calculus provides stochastic counterparts to derivatives and integrals for functions of stochastic processes. 2) Itô's lemma gives the stochastic version of the chain rule for functions of stochastic processes like stock prices that follow stochastic differential equations (SDEs). 3) Applying Itô's lemma, the SDE for a derivative can be derived once the SDE of the underlying asset is known. Itô's formula is thus useful for finding SDEs and evaluating stochastic integrals.

Uploaded by

Norbert Durand
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 25

5.

Itô Calculus

Types of derivatives

Consider a function F (St, t) depending on two


variables St (say, price) and time t, where
variable St itself varies with time t.

In standard calculus there are three types of


derivatives:

Partial derivative:
∂F (St, t) ∂F (St, t)
(1) Fs = , Ft = .
∂St ∂t
Total derivative:

(2) dF = Fs dSt + Ft dt.


Chain rule:
dF (St, t) dSt
(3) = Fs + Ft.
dt dt

1
Partial derivative are abstractions. Usually
they are called multipliers or marginal effects
(cf. the Greeks in option theory).

Total derivative describes the total change or


response in F as time t and St change

The chain rule indicates the chain effects in


the change of the price St and F as time
changes.

We will consider in this section stochastic


counterparts for the total differential and chain
rule. Essentially as we will see the major dif-
ferences are that we have to interpret the dif-
ferential in stochastic processes via the stochas-
tic integral and that the second order term
(dSt)2 is not negligible as in standard calcu-
lus.

2
Itô’s stochastic differential equation

Itô’s lemma gives the stochastic version for


the chain rule.

Let

(4) dSt = a(St, t) dt + σ(St, t) dWt.


where a(St, t) and σ(St, t) are nonanticipating
and Wt is the standard Brownian motion.

3
We interpret dSt via the stochastic integral
such that
Z t
(5) dSu = St − S0,
0
so that

Z t Z t Z t
(6) St = S0 + dSu = a(Su , u) du + σ(Su , u) dWu ,
0 0 0

where the first integral is the usual Riemann


integral and the second one is the Itô inte-
gral.

4
Consider a function F (St, t) (e.g. derivative
of a stock)

As time t changes by dt what is the total


effect on F (St, t). The change is

(7) t → Wt → St → F (St, t).


The interest is dF (St, t).

5
Suppose the observation interval of St is [0, T ].
Let 0 = t0 < t1 < · · · < tn = T be a partition
with
T
(8) h = tk − tk−1 = k = 1, . . . , n,
n
so that T = nh.

Consider the finite difference representation


of dSt

(9) ∆Sk = ak h + σk ∆Wk , k = 1, . . . , n,


with ∆Sk = Stk −Stk−1 , ak = a(Stk−1 , tk ), σk =
σ(Sk−1, tk ), and ∆Wk = Wtk − Wtk−1 .

6
Itô formula is derived using the Taylor expan-
sion of a ”smooth” function.

If f (x) is such a function the Taylor expansion


around x0 becomes

1 00
f (x) = f (x0) + f 0 (x0)(x − x0 ) + f (x0 )(x − x0 )2 + R,
2
(10)

where R is the remainder.

7
For a function with two variables the Taylor
expansion is
F (Stk , tk ) = F (Stk−1 , tk ) + Fs ∆Sk + Fth + 12 Fss (∆Sk )2

+ 12 Ftt h2 + Fst h ∆Sk + R.


(11)

Arranging terms
∆Fk = Fs∆Sk + Fth + 2 1 F (∆S )2
ss k
(12)
+12 F tt h2 + F h ∆S + R,
st k
where

(13) ∆Fk = F (Stk , tk ) − F (Stk−1 , tk−1).

8
As n → ∞, h = tk −tk−1 → dt, ∆Sk → dS, and
∆Fk → dF , and R → 0 because it consists of
terms (∆tk )m and (∆Wk )m with m ≥ 3.

So we get
dF (St, t) = Fs dSt + Ft dt + 12 Fss (dSt )2
(14)
+ 12 Ftt (dt)2 + Fst dt dSt.

Using the calculation rules for the differen-


tials, we obtain
dt dSt = dt (a(St, t)dt + σ(St, t) dWt)

(15) = a(St, t)(dt)2 + σ(St, t) dt dWt

= 0,
because (dt)2 = 0 and dt dWt = 0.

So we get
1
(16) dF (St, t) = Fs dSt + Ft dt + Fss (dSt)2
2

9
Remark 5.1: If St is non-stochastic then (dSt )2 = 0
and the above formula is just the total derivative dF =
Fs dS + Ft dt.

Replacing dSt with its Itô representation, we


have
dF (St, t) = Fs · [a(St , t) dt + σ(St , t) dWt ] + Ft dt
(17)
+ 21 Fss [a(St, t) dt + σ(St, t) dWt]2 .
Using the infinitesimal calculation rules again
yields
(18) (dSt)2 = σ 2(St, t) dt.
Arranging terms, we obtain finally the fa-
mous Itô’s differential formula

· ¸
1 2
dF = Ft + a(St , t)Ft + σ (St , t)Fss dt + σ(St, t) dWt.
2
(19)

10
The result is summarized as Itô’s Lemma:

Lemma 5.1: (Itô’s Lemma) Let F (St , t) be a twice-


differentiable function of t and of the random process
St with Itô differential equation
dSt = at dt + σt dWt , t ≥ 0,
with at = a(St, t) and σt = σ(St , t) continuously twice-
differentiable (real valued) functions. Then
1
(20) dF = Fs dSt + Ft dt + Fss σt2 dt,
2
or, after substituting for the right hand side of dSt
above
µ ¶
1
(21) dF = Fs at + Ft + Fss σ 2 dt + Fs σt dWt ,
2
where
∂F ∂F ∂ 2F
(22) Fs = , Ft = , and Fss = .
∂St ∂t ∂St2

11
The major usage of the Itô formula in fi-
nance is to find the (Itô) stochastic differen-
tial equation (SDE) for the financial deriva-
tive once the (Itô) SDE of the underlying
asset is given.

12
Itô’s formula can be used also in some cases
to find the stochastic integral itself.

Example 5.1: Let

(23) F (Wt, t) = Wt2 .


Using formula (16) with St = Wt we obtain
(24) Fw = ∂W 2 /∂W = 2W ,
and
(25) Fww = ∂ 2F/∂W 2 = 2.
Then because Ft = 0,
dF (Wt , t) = Fw dWt + 12 Fww (dWt )2
(26)
= dt + 2Wt dWt.
Thus the drift of F is a(F, t) = 1 and diffusion param-
eter is σ(F, t) = 2Wt .

13
Example 5.2:

(27) F (Wt , t) = 3 + t + eWt .


Using again Itô’s formula (16) with St = Wt
dF (Wt, t) = Ft dt + Fw dWt + 21 Fww (dWt)2

(28) = dt + eWt dWt + 12 eWt dt


¡ ¢
= 1 + 21 eWt dt + eWt dWt .

14
Example 5.3: Consider the geometric Brownian mo-
tion
1
St = S0e{(µ− 2 σ )t+σ Wt }
2
(29) ,
where S0 is a constant. Then using again Itô with
formula (20), and noting that σ(Wt, t) = 1, we get
∂St ∂St 1 ∂ 2 St
dSt = ∂Wt
dWt + ∂t
dt + 2 ∂Wt2
dt
1 1
= S0 σe{(µ− 2 σ )t+σ Wt }
dWt + (µ − 12 σ 2 )S0 e(µ− 2 σ )t+σ Wt
2 2
dt
1
+ 12 σ 2 S0e(µ− 2 σ )t+σ Wt
2
dt

= Stσ dWt + (µ − 12 σ 2 )St dt + 12 σ 2 St dt

= St(µ dt + σ dWt),
(30)
or
dSt
(31) = µ dt + σdWt ,
St
or
(32) dSt = µStdt + σSt dWt .

Remark 5.2: Comparing to the general formula dSt =


a(St, t)dt + σ(St , t)dWt we find that in (32) a(St, t) =
µ St, and σ(St, t) = σ St.

15
Itô’s formula as an integration tool

Suppose our task is to evaluate


Z t
(33) Ws dWs.
0
Make a guess
1 2
(34) F (Wt, t) = Wt .
2
Then using Itô
1
(35) dF (Wt, t) = Wt dWt + dt.
2
The integral form is
Z t Z t Z t
1 2 1
W = F (Wt , t) = dF (Ws , s) = Ws dWs + ds.
2 t 0 0 2 0
(36)
So
Z t
1 2 1
(37) Ws dWs =Wt − t.
0 2 2
The start off point here is to make a “good
guess”.

16
Example 5.4: Consider Itô integral
Z t
(38) s dWs .
0
Make a start off guess
(39) F (Wt, t) = tWt.
Then
(40) dF (Wt, t) = Wt dt + t dWt.
and
Z t Z t Z t
(41) tWt = dF (Ws , s) = Ws ds + s dWs .
0 0 0
So
Z t Z t
(42) s dWs = tWt − Ws ds.
0 0

17
Example 5.5: Consider
dSt
(43) = µ dt + σ dWt.
St
Let
(44) F (St , t) = log St.
Then
dF (St, t) = Ft dt + Fs dSt + 21 Fss (dSt)2

1 1 1
= St
dSt − 2 St2
(dSt )2
(45)
1 1 2 2
= µ dt + σ dWt − 2 St2
σ St dt

= (µ − 12 σ 2 ) dt + σ dWt.
We get
Rt
log St = log S0 + 0
dF (Su , u)
Rt 1 2
Rt
(46) = log S0 + 0
(µ − 2
σ )du + 0
σ dWu

= log S0 + (µ − 21 σ 2)t + σWt .


So
1
St = S0 e(µ− 2 σ )t+σWt
2
(47) .

18
Integral form of Itô’s Lemma

Integrating both sides of (21) yields Itô’s for-


mula in integral form:
Z t· ¸
1
F (St, t) = F (S0 , 0) + F2 (Su , u) + F11(Su , u)σu2 du
0 2
Z t
+ Fs dSu .
0
(48)
where
∂F (x, y) ∂F (x, y)
(49) F1 (x, y) = ,F2(x, y) = ,
∂x ∂y
and
∂ 2F (x, y)
(50) F11(x, y) = ,
∂x2
and we have used
Z t
(51) dF (Su, u) = F (St, t) − F (S0, 0).
0

19
Remark 5.3: Rearranging terms in the Itô’s integral form yields
Z t
Fs dSu = [F (St , t) − F (S0 , 0)]
0
(52) Z th i
1
− F2 (Su , u) + F11 (Su , u)σu2 du,
0
2

which is a representation of the stochastic integral as a function

of integrals with respect to time.

20
Multivariate Itô formula

(53)
µ ¶ µ ¶ µ ¶µ ¶
dS1 (t) a1 (t) σ11 (t) σ12 (t) dW1 (t)
= dt+
dS2 (t) a2 (t) σ21 (t) σ22 (t) dW2 (t)
or
(54)
dS1 (t) = a1(t) dt + σ11(t) dW1 (t) + σ12 (t) dW2 (t)
dS2 (t) = a2(t) dt + σ21(t) dW1 (t) + σ22 (t) dW2 (t),
where it is assumed that Wiener processes
W1(t) and W2(t) are independent.

21
Suppose F (S1(t), S2(t), t) is a twice differen-
tiable real valued function.

Use of the Taylor expansion and taking limit


in the same manner as in the univariate case,
yields (with (dt)2 = 0, dt dS1 = 0, and dt dS2 =
0)

(55)
dF = Ft dt + Fs1 dS1 + Fs2 dS2
£ ¤
+ 12 2 2
Fs1 ,s1 (dS1 ) + Fs2 ,s2 (dS2 ) + 2Fs1,s2 dS1 dS2 .

The independence of W1 and W2 implies that


dW1 dW2 = 0 (otherwise if they were cor-
related with correlation ρ, then dW1 dW2 =
ρ dt).

22
Then
(56) (dS1)2 = (σ11
2 + σ 2 ) dt,
12

(57) (dS2)2 = (σ21


2 + σ 2 ) dt,
22
and

(58) dS1 dS2 = (σ11σ21 + σ12σ22)dt.

Using these in the multivariate Itô gives a


formula as a function of dW1 and dW2.

23
Example 5.6: Interest rate derivatives. Assume that
the yield curve depends on two state variables, short
rate rt, and long rate Rt . Denote the price of the
derivative as F (rt, Rt, t). Assume
(59) drt = a1 (t) dt + σ11(t) dW1 (t) + σ12 (t) dW2 (t),
and
(60) dRt = a2 (t) dt + σ21 (t) dW1 (t) + σ22 (t)dW2(t).
Straightforward application of the Itô formula gives
(61)

dF = Ft dt + Fr drt + FR dRt
£ ¤
+ 12 Frr (σ11
2 + σ2 ) + F
12
2 2
RR (σ21 + σ22 ) + 2FrR (σ11 σ21 + σ12 σ22 ) dt,

which indicates how the price of an interest rate deriva-


tive will change during a small interval dt.

Remark 5.4:
(62) Cov(drt , dRt ) = [σ11 (t)σ21 (t) + σ12 (t)σ22 (t)] dt.

24
Example 5.7: Total value of wealth
n
X
(63) Y (t) = Ni (t)Pi (t),
i=1

where Ni (t) is units of the ith asset and Pi (t) the price.
Increment of wealth as time passes
Xn Xn
∂Y ∂Y ∂Y
dY (t) = dt + dNi (t) + dPi (t)
∂t i=1
∂Ni i=1
∂Pi
n n
1 X ∂ 2Y 2 1 X ∂ 2Y 2
+ (dN i (t)) + (dP i (t))
2 i=1 ∂Ni2 2 i=1 ∂Pi2
Xn
∂ 2Y
+ dNi (t) dPi (t)
i=1
∂N i ∂P i

n
X n
X
= Pi (t)dNi (t) + Ni (t)dPi (t)
i=1 i=1
n
X
+ dNi (t) dPi (t).
i=1
(64)
In the standard calculations the last term would not
be present.

25

You might also like