Midterm solutions, Spring 2025 1
1. Consider three independent Brownian motions W 1 , W 2 , and W 3 . Two risky assets evolve according
to the Black-Scholes dynamics:
dSt1 dSt2
= µ1 dt + σ1 dWt1 , = µ2 dt + σ2 dWt2 ,
St1 St2
while the short rate process is given by
q
Rt = µR dt + σR ρ1 Wt1 + ρ2 Wt2 + 1 − ρ21 − ρ22 Wt3 ,
where ρ1 and ρ2 are constants in (−1, 1) satisfying ρ21 + ρ22 < 1.
(a) Are there arbitrage opportunities in the above model?
Solution. No. Under an equivalent risk-neutral measure Q, the drift of each risky asset must equal
the short rate Rt . Thus, if we denote the market price of risk (MPR) processes by θ1 , θ2 , and θ3
(corresponding to the three independent Brownian motions), the change of measure given by
( Z )
t Z t
dQ 1
θ1 dWs1 + θ2 dWs2 + θ3 dWs3 − θ2 + θ22 + θ32 ds ,
= exp −
dP Ft 0 2 0 1
must be such that the discounted asset prices become Q-martingales.
Note that the MPR equations can be solved uniquely for θ1 and θ2 :
µ1 − Rt µ2 − Rt
θ1 = , θ2 = .
σ1 σ2
Since these equations have solutions, by the first FTAP, the model is arbitrage free.
(b) Is the market model complete?
Solution. No. Because W 3 does not influence the dynamics of any traded asset, the corresponding
market price of risk θ3 is completely free (i.e. it is not pinned down by any hedging requirement).
In other words, one can choose any (sufficiently integrable) process for θ3 and still have a valid
Radon-Nikodym derivative, leading to many possible risk-neutral measures. Thus, by the second
FTAP, it is incomplete.
2. (a) Prove put-call parity.
Proof. Consider a forward contract with delivery price K whose value at expiration is S(T ) − K.
Let f (t, x) denote the value of the forward contract at time t when S(t) = x; by a no-arbitrage
replication argument, one may replicate the forward payoff by buying one share of stock and
financing it by borrowing e−r(T −t) K from the money market, so that f (t, x) = x − e−r(T −t) K.
Since for any number x the identity
x − K = (x − K)+ − (K − x)+
holds with (x − K)+ being the call payoff and (K − x)+ the put payoff, the forward payoff at
expiration satisfies
S(T ) − K = c(T, S(T )) − p(T, S(T )).
Midterm solutions, Spring 2025 2
By no-arbitrage, the value of the forward contract must equal the value of a portfolio that is long
a call and short a put, so that for all t we have
f (t, S(t)) = c(t, S(t)) − p(t, S(t)).
Substituting the replication result yields the put-call parity formula:
c(t, S(t)) − p(t, S(t)) = S(t) − e−r(T −t) K.
One can check out more detailed discussion in [Shr04, §4.5.6].
(b) Calculate out the volatility function σ for exact calibration using Dupire’s formula.
Proof. The assumptions for the put option price given in the problem is equivalent to using the
call price
2
ln(S0 /K) + r ± c2 T
c(0, S0 ; T, K) = S0 N (d+ ) − Ke−rT N (d− ), d± = √
c T
where c is some positive constant.
First note that
N ′ (d+ )
2
d+ − d2−
(d+ + d− )(d+ − d− ) K −rT
= exp − = exp − = e . (⋆)
N ′ (d− ) 2 2 S0
Then, by the chain-rule, one obtains
cT (0, S0 ; T, K) = S0 N ′ (d+ )∂T d+ + rKe−rT N (d− ) − Ke−rT N ′ (d− )∂T d−
(⋆)
= Ke−rT N ′ (d− )∂T (d+ − d− ) + rKe−rT N (d− )
1
= √ cKe−rT N ′ (d− ) + rKe−rT N (d− ).
2 T
√
Next, noting that ∂K d± = −1/ K c T , one obtains
∂
S0 N (d+ ) − Ke−rT N (d− )
cK (0, S0 ; T, K) =
∂K
S0 N ′ (d+ ) N ′ (d− ) (⋆)
=− √ − e−rT N (d− ) + e−rT √ = −e−rT N (d− ).
Kc T c T
Finally, differentiating cK with respect to K yields
∂ −rT ∂d− e−rT N ′ (d− )
N (d− ) = −e−rT N ′ (d− )
cKK (0, S0 ; T, K) = −e = √ .
∂K ∂K Kc T
Plugging in everything, Dupire’s formula yields
2 (cT (0, S(0); T, K) + rKcK (0, S(0); T, K))
σ(T, K)2 =
K 2 cKK (0, S(0); T, K)
,
e−rT ′ (d
1 −rT ′ 2 N −)
= √ cK e N (d− ) K √ = c2
T K cT
Taking the positive solution yields
σ(T, K) = c,
a constant.
Midterm solutions, Spring 2025 3
3. (20 points) We wish to compute the conditional expectation
h i
e (W
E f (T ) + T )2 e−W
f (T )
| F(t) (⋆⋆)
where W
f is a standard Brownian motion under the probability measure P.
e
(a) Find a measure P such that W
f (t) + t is a Brownian motion under P.
Solution. It is straightforward that, if we define
dP T
ZT := T
= exp −WT − ,
dP
e 2
then under P the process
Bt := Wt + t, 0≤t≤T
is a standard Brownian motion.
(b) Compute the conditional expectation in (⋆⋆).
Solution. Note that by Bayes’ rule, for any FT -measurable random variable X,
e T X | Ft ]
E[Z
E[X | Ft ] = , =⇒ e T X | Ft ] = Zt · E[X | Ft ].
E[Z
Zt
Let
X := eT /2 (WT + T )2 = eT /2 (BT )2
Then, noting that
E[(BT )2 | Ft ] = E (Bt + (BT − Bt ))2 | Ft
= (Bt )2 + 2Bt E[BT − Bt ] + E[(BT − Bt )2 ] = (Bt )2 + T − t,
one has
e e−WT (WT + T )2 Ft = E[Z
E e T X | Ft ]
= Zt · E[X | Ft ]
= e−Wt −t/2 · E[eT /2 (BT )2 | Ft ]
= e−Wt −t/2 · eT /2 E[(BT )2 | Ft ]
= e−Wt −t/2 · eT /2 (Bt )2 + T − t
T −t
(Wt + t)2 + T − t .
= exp −Wt +
2
References
[Shr04] Steven E. Shreve. Stochastic Calculus for Finance II: Continuous-Time Models. Springer, New York,
NY; Heidelberg, 2004. doi:10.1007/978-0-387-40101-6.
