Exercise Class No. 7 - Financial Mathematics I

Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

EXERCISE CLASS NO.

7 — FINANCIAL MATHEMATICS I

PROFESSOR DR PETER BANK (LECTURER),


FRANZISKA BIELERT (ASSISTANT)

This is the material for the exercise classes on 5 and 8 December 2022.

1. Numéraire Invariance
Let S̄ denote a T -period market model with d risky assets defined on a filtered
probabilty space (Ω, A , F , P) with P-trivial F0 . Suppose that P(St1 > 0) = 1 for all
t = 0, . . . , T . We can thus consider S 1 as a numéraire as well. Denote by

iS0 i
X̃ = 1 X , i = 0, . . . , d
S
the discounted price processes with respect to the numéraire S 1 . Recall that P is
the set of equivalent martingale measures with respect to the numéraire S 0 , that
is: P∗ ∈ P means that P∗ is an equivalent probability measure to P, and X i is a
P∗ -martingale for all i = 0, . . . , d.
Let P̃ be the set of equivalent martingale measures with respect to the numéraire

S , that is P̃ ∈ P̃ iff the preceding condition holds true for X (resp. P∗ ) replaced
1

with X̃ (resp. P̃ ). Further let C be a contingent claim with maturity T . Denote
by Π(C/ST0 ) (resp. Π̃(C/ST1 )) the set of arbitrage-free prices with respect to the
numéraire S 0 (resp. S 1 ), quoted in units of S00 (resp. S01 ).
(i) Argue that P is empty (is a singleton) if and only if P̃ is, respectively.

(ii) Show that P → P̃, P∗ 7→ P̃ , defined by

dP̃ S00 ST1
=
d P∗ S01 ST0

is a bijection.
(iii) Show that, for any contingent claim C with maturity T ,

Π((CS00 )/ST0 ) = Π̃((CS01 )/ST1 ).

Date: December 1, 2022.


1
2 PROFESSOR DR PETER BANK (LECTURER), FRANZISKA BIELERT (ASSISTANT)

2. La martingale: Doubling or Ruin I


Consider the following binomial model with infinite time horizon. Let Ω = {a, b}N
for some a, b ∈ R with −1 < a < b, A = P({a, b})⊗N and R be the canonical process
on Ω (i.e., Rt (ω) = ωt ), further r > −1. Then the financial market model is given by
t
St0 = (1 + r)t , St1 =
Y
(1 + Rs ),
s=1
1
the filtration F = F R = F S and a probability measure P, such that (Rt )t∈N is an
iid sequence of Bernoulli random variables for parameter p ∈ (0, 1). More precisely,
we set for all finite, disjoint Ta , Tb ⊂ N,
P({ω ∈ Ω | ∀t ∈ Ta : ωt = a, ∀t ∈ Tb : ωt = b}) = (1 − p)#Ta p#Tb .
Suppose that r ∈ (a, b) and recall from the lecture that this implies that the absence
of arbitrage in finite time.
A trading strategy in discrete infinite time is simply a predictable process ξ¯ = (ξ 0 , ξ 1 )
that is indexed over N. Writing X̄ = S̄/S 0 = (1, X) for the discounted prices, the
¯ ¯
corresponding wealth process is given by V0ξ = ξ¯1 · X̄0 and Vtξ = ξ¯t · X̄t at time t ∈ N
and the strategy is self-financing iff for all t ∈ N,
(ξ¯t+1 − ξ¯t ) · S̄t = 0.
Exactly as in finite time multi-period models, there is for every predictable process
¯
ξ 1 a unique predictable ξ 0 , such that ξ¯ = (ξ 0 , ξ 1 ) is self-financing with V0ξ = 0, and
¯ it holds that
for every self-financing ξ,
¯ ¯
Vtξ = Vt−1
ξ
+ ξt1 (Xt − Xt−1 ).
1+r
i) Let γ ∈ R be such that γ > 1+a . We recursively define a self-financing trading
¯
ξ
strategy ξ¯ with V = 0 by setting ξ 1 = 1 and for t ∈ N,
0 1
1
ξt+1 = γξt1 1Rt =a .
Show that for all t ∈ N, ω ∈ Ω with R1 (ω) = · · · = Rt (ω) = a, it holds with
δ = γ 1+a
1+r , that
ξ¯ a − r δt − 1
Vt (ω) = .
1+r δ−1
ii) Show that for all t ∈ N, ω ∈ Ω with R1 (ω) = · · · = Rt (ω) = a and Rt+1 (ω) = b, it
holds
ξ¯ a − r δt − 1 b−r
Vt+1 (ω) = + δt .
1+r δ−1 1+r
iii) Show that the preceding right-hand side term is strictly positive for all t ≥ 0 iff
b−a b−a 1+r
δ≥ , alias γ ≥ . (1)
b−r b−r 1+a
EXERCISE CLASS NO. 7 — FINANCIAL MATHEMATICS I 3

¯
iv) Interpret the strategy ξ.
3. La martingale: Doubling or Ruin II
Assume the setting from Exercise 2 and condition (1). We consider
τ = inf{t ∈ N : Rt = b}.
i) Show that τ is P-a.s. finite, i.e., P(τ < ∞) = 1.
¯
ii) Show that Vτξ > 0, P-a.s.
iii) Show that there is a unique p∗ ∈ (0, 1) inducing P∗ , such that E∗ R1 = r. Provide
a formula for p∗ in terms of a, r, b, and show that X is a martingale with respect
to P∗ .
iv) Show that τ is geometrically distributed under P∗ for somce success probability
π ∈ (0, 1), i.e., for t ∈ N
P∗ (τ = t) = π(1 − π)t−1 .
¯
v) Show that E∗ Vτξ−1 = −∞.
4. La martingale: Doubling or Ruin III
In the setting of the previous two problems, we conclude that there is an equivalent
martingale measure, a self-financing trading strategy ϕ̄ and an P-almost surely finite
stopping time σ, such that
(∗) V0ϕ̄ = 0, P(Vσϕ̄ ≥ 0) = 1, P(Vσϕ̄ > 0) > 0. (2)
(i) Show that (2) is impossible for any pair (ϕ̄, σ) of a self-financing trading strategy
and an P-almost surely finite stopping time such that additionally, there is some
C ∈ R, such that P-almost surely for all t ∈ N
Vtϕ̄ ≥ C. (3)
(ii) Argue in which sense ξ¯ is an arbitrage opportunity or not. What is the financial
meaning of condition (3)?
Hint: Establish the supermartingale property of the discounted wealth process.
You are allowed to apply the optional sampling theorem.
Technische Universität Berlin, Fakultät II, Institut für Mathematik, Arbeits-
gruppe Stochastik und Finanzmathematik, Sekr. 7–1, Straße des 17. Juni 136, D-
10623 Berlin
Email address: [email protected]

You might also like