MATH97133 Advanced Statistical Finance Mock Exam
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
MATH97133
Advanced Statistical Finance
Mock Exam
1. (a) Let X = (Xt )t>0 be a continuous-time stochastic process. Define the quadratic variation [X]t of X at time t 2 0.
(b) Suppose that X is the deterministic process Xt = et , t 2 0. Show, using the definition in (a), that [X]t = 0 for any t 2 0. (You may quote that 0 < ex - ey < ex (x - y) for x 2 y.)
(c) Let X now be an Itˆo process
t t
Xt = X0 + µu du + σu dWu , t 2 0,
0 0
where W = (Wt )t>0 is a standard Brownian motion, and let (t, x) 1- f (t, x) be once continuously differentiable in t and twice continuously differentiable in x. State Itˆo’s formula for f (t, Xt ) for any t 2 0 starting from f (0, X0 ). Specify also the type of each integral appearing in the formula.
(d) Suppose that X satisfies a stochastic differential equation
dXt = cos(Xt )dWt , t 2 0,
where W is a standard Brownian motion. Show that Yt := Xt(2) , t 2 0, is an Itˆo process and determine [Y]t for any t 2 0.
2. (a) Let ~1 , . . . , ~T be d-dimensional (d 2 2) iid random vectors. Show that the sample
covariance matrix
T
Σd,T := ╱~t - ~、╱~t - ~、/ ,
t=1
where ~ := ~t is the sample mean vector, is invertible only if T 2 d. (b) Let X be a random variable with moment generating function (mgf) ϕX (t) := r[etX ],
t e 冈. Show that o
ϕX (t) = t ut-1 FX (log u)du, t > 0,
0
where FX is the cdf of X and FX (x) := 1 - F (x).
(c) Keeping the notation of (b), suppose that X follows now the Pareto distribution with tail index α > 0 and scale κ > 0, so that
FX (x) =
Determine for which t e 冈 the mgf ϕX (t) is finite. μψ≠1妓 For t > 0, it is useful to note that u 1- FX (log u) is slowly varying.
3. (a) Explain briefly what market microstructure noise is and what its main sources are.
Consider the following stylised model of market microstructure noise. The observed logarithmic price log S at time , for any i = 0, 1, . . . , n and n e 日, is given by
log S = log + εi ,
where log is the latent logarithmic price at time and εi is a market microstructure noise
term. We assume that
log t = σWt , t e [0, 1],
where σ > 0 is a constant and W = (Wt )te「0 ,1| is a standard Brownian motion. Additionally, we assume that εi , i = 0, 1, . . . , n, are iid with ε0 ~ N(0, v2 ), for some constant v > 0, and independent of W .
(b) The realised variance corresponding to the observed logarithmic prices log S n , i =
_
0, 1, . . . , n, is defined by
n
RVn := (log S - log S )2 .
i=1
Show that r[RVn] = σ 2 + 2nv2 . What can you conclude concerning the impact of market microstructure noise on RVn and its asymptotic behaviour as n - o?
(c) Define the h-th (h = 1, . . . , n - 1) realised autocovariance
n
γn (h) := (log S - log S )(log S - log S ).
i=h+1
Let H = 1, . . . , n - 1. Determine all weights k1 , . . . , kH e 冈 that make the realised
kernel
H
RKn,H := RVn + kh γn (h)
h=1
is an unbiased estimator of σ 2 .
2022-05-11