CQF Module 1.
3 Exercises
Throughout this problem sheet, you may assume that Wt is a Brownian Motion (Weiner Process) and dWt
is its increment; and W0 = 0:
1. Use Itô’s lemma to obtain a SDE for each of the following functions
(a) y (Wt ) = exp (Wt )
(b) g (Wt ) = ln Wt
(c) h (Wt ) = sin Wt + cos Wt
(d) f (Wt ) = aWt ; where the constant a > 1
(e) f (Wt ) = (Wt )n
2. Using the formula below for stochastic integrals, for a function F (Wt ; t) ;
Z t Z t
@F @F 1 @2F
dWt = F (Wt ; t) F (W0 ; 0) + d
0 @Wt 0 @ 2 @W 2
show that we can write
Z t Z t
a. Wt3 dW = 41 W 4 (t) 3
2
W 2d
0 0
Z t Z t
b. dW = tWt W d
0 0
Z t Z t
1 1
c. (W + ) dW = 2
Wt2 + tWt W + 2
d
0 0
1
CQF Module 1.4 Exercises
Stochastic Di¤erential Equations and Itô’s Lemma
Throughout this problem sheet, you may assume that Wt or W (t) is a
Brownian Motion (Weiner Process) and dWt (or dW (t)) is its increment. W0 =
0:
1. The change in a share price S (t) satis…es
dS = A (S; t) dWt + B (S; t) dt;
for some functions A and B: If f = f (S; t) ; then Itô’s lemma gives the
following stochastic di¤erential equation
@f @f 1 @2f @f
df = +B + A2 2 dt + A dWt :
@t @S 2 @S @S
Can A and B be chosen so that a function g = g (S) has a change which
has zero drift, but non-zero di¤usion? State any appropriate conditions.
2. Show that F (Wt ) = arcsin (2aWt + sin F0 ) is a solution of the stochastic
di¤erential equation
dF = 2a2 (tan F ) sec2 F dt + 2a (sec F ) dWt ;
where F0 and a is a constant.
3. Show that
Z t Z t
W 2( )
W( ) 1 e dW ( ) = F (W (t)) + G (W (t)) d :
0 0
where the functions F and G should be determined.
4. Consider a two factor model in which the stock price dynamics St ; follows
Geometric Brownian Motion and the stock variance vt is itself stochastic
and follows a square root process
p
dSt = St dt + vt St dW1 (t);
p
dvt = (vt v)dt + vt dW2 (t):
The two processes have a correlation coe¢ cient , i.e.
dW1 (t)dW2 (t) = dt
The parameters ; ; v and are all constant. Let F = F (t; St ; vt ):
Using Itô, consider the SDE for dF and integrate over [0; t] to obtain an
expression for F (t; St ; vt ):
1
Module 1 Further Exercises in SDEs 2015
Throughout this problem sheet, you may assume that Wt is a Brownian Motion (Weiner Process) and dWt
is its increment. You may assumeW0 = 0: SDE(s) refers to Stochastic Di¤erential Equation(s).
1. Let be a random variable which follows
p a standardised normal distribution, i.e. N (0; 1) :
Calculate E [ ] and V [ ] where = dt : dt is a small time-step. Note: No integration is
required.
2. Consider the following examples of Stochastic Di¤erential Equations (SDE); Write these in standard
form, i.e. dG = A(G; t)dt + B(G; t)dWt . Give the drift and di¤usion for each case.
p
(a) df + dWt dt + 2 tf dt + 2 f dWt = 0 where f = f (Wt ; t)
dy
(b) = (A + By) dt + (Cy) dWt where y = y (Wt ; t)
y
(c) dS = ( S)dt + dWt + 4dS
3. Show that Z Z
1 1
(1 t) cos Wt dWt = (a + bt) sin Wt dt;
0 0
and determine the values of a and b:
4. The function V (S; t) = log (tS), where S evolves according to the SDE dS = Sdt + SdWt ; show
that
1 1 2
dV = + dt + dX:
t 2
5. Show that
G = exp t + aeWt
is a solution of the stochastic di¤erential equation
dG (t) = G 1 + 21 (ln G t) + 21 (ln G t)2 dt + G (ln G t) dWt
6. Consider the stochastic di¤erential equation
dG (t) = a (G; t) dt + b (G; t) dWt :
Find a (G; t) and b (G; t) where
(a) G (t) = Wt2
(b) G (t) = 1 + t + eWt
(c) G (t) = ft Wt ; where ft is a bounded and continuous function.
7. Use Itô’s lemma to show that
d (cos Wt ) = (cos Wt ) dt + (sin Wt ) dWt
&
d (sin Wt ) = (sin Wt ) dt (cos Wt ) dWt
and determine the constants & :
1
CQF Module 2, Session 5: Martingales I
Exercises
CQF
1. Use Itô’s formula to determine whether the following are martingales:
(i) Y (t) = e1/2t cos X(t);
(ii) Y (t) = eαt sin X(t) for some constant α with 0 < α < 1. Does
the answer depend on the value of α?
(iii) Y (t) = (X(t) + t) exp − 21 t − X(t) .
2. Moments of the Brownian Motion X(t) - Consider the function
mn (t) defined as
mn (t) = E[X n (t)], n = 1, 2, . . . (1)
where X(t) is a standard Brownian motion.
Applying Itô’s formula, show that:
Z t
1
mn (t) = n(n − 1) mn−2 (s)ds (2)
2 0
for n = 2, 3, . . .
Deduce from (2) that
m4 (t) = 3t2 (3)
compute m6 (t).
3. Let Xn , n = 1, . . . be i.i.d random variables where P (Xn = 1) = p and
P (Xn = −1) = 1 − p. You can think of Xn as being the nth coin
toss in a sequence. Let Sn , n = 1, . . . be the associated random walk,
defined as
Sn = X1 + X2 + . . . + Xn (4)
1
Sn can be viewed as the P&L of the entire coin toss game. We also
introduce the filtration Fn generated by the Xn and such that Xn is
Fn -adapted.
Find conditions under which the random walk is (a) a martingale, (b)
a submartingale (c) a supermartingale.
4. Let Yt = Xt4 where Xt is a Brownian motion. Using Itô’s lemma, ex-
press the SDE for Yt . Then, deduce the stochastic integral for Yt over
[0, T ]. Finally, deduce from the stochastic integral an expression for
E[Yt ].
5. Discrete Time Martingale: Let Y1 , . . . , Yn be a sequence of inde-
pendent random variables such that E[Yi ] = 0 for i = 1, . . . , n. Let Fn
be the filtration generated
P by the sequence Y1 , . . . , Yn . Consider the
random variable Sn = ni=1 Yi . Prove that Sn is a martingale for all n.
Reminder - proving that a process Sn is a martingale involves proving
that E[|Sn |] < ∞ and that E[Sn+1 |Fn ] = Sn
