2
$\begingroup$

I'm trying to prove that for a Brownian motion $\big(B_t, \mathcal{F}_t \big)_{t\geq 0}$ and a stopping time $\tau$ satifying $\mathbb{E}[\tau]<\infty$, we have that $\mathbb{E}[B_\tau^2]=\mathbb{E}[\tau]$.

I know that $U_t=W_t^2-t$ is a martingale, and It's enough for me to show that $\mathbb{E} \big[ U_\tau\big]=0 $. I define a sequence $U_{\tau \wedge n}$, and because it's a martingale I know by Doob's O.S.T that $\mathbb{E}[U_{\tau \wedge n}]=0$. Since $U_{\tau \wedge n}\rightarrow U_\tau$ almost surely, if I find a dominating function for $U_{\tau \wedge n}$, I will obtain that:

$0\equiv\mathbb{E}[U_{\tau \wedge n}]\rightarrow \mathbb{E}[U_\tau]$, which solves the problem.

However I've been having problems finding a dominating function for $U_{\tau \wedge n}$. I do know that:

$$\vert U_{\tau \wedge n}\vert \leq B_{\tau \wedge n}^2+\tau= B_\tau^2\cdot 1_{ \{ \tau \leq n\}}+ B_n^2\cdot 1_{ \{ \tau > n\}}+\tau \leq B_\tau^2+ B_n^2\cdot 1_{ \{ \tau > n\}}+ \tau$$

By Fatou's lemma I know that $B_\tau^2\in L^1$. Hence I already bounded the first and last term in $L^1$, and I just need to find an $L^1$ bound on $B_n^2\cdot 1_{ \{ \tau > n\}}$ to conclude.

I would very much appreciate any hint, as it seems to me as there is a little thing I am missing.

$\endgroup$
3
$\begingroup$

Without any additional information on the stopping time it will be difficult to find an integrable dominating function, I think. In order to avoid the problem of finding a dominating function you can use the following alternative approach:

Use the martingale property of $(B_{t \wedge \tau})_{t \geq 0}$ and $(U_{t \wedge \tau})_{t \geq 0}$ to show that

$$\mathbb{E}((B_{n \wedge \tau}-B_{m \wedge \tau})^2) = \mathbb{E}(\tau \wedge n)-\mathbb{E}(\tau \wedge m), \qquad m \leq n,$$

and conclude that $(B_{n \wedge \tau})_{n \in \mathbb{N}}$ is an $L^2$-Cauchy sequence. By the completeness of $L^2$, the limit $X := \lim_{n \to \infty} B_{n \wedge \tau}$ exists in $L^2$. Since you already know that the sequence converges almost surely to $B_{\tau}$, it follows that $X=B_{\tau}$, and so $B_{\tau \wedge n} \to B_{\tau}$ in $L^2$. In particular,

$$\mathbb{E}(B_{\tau \wedge n}^2) \xrightarrow[]{n \to \infty} \mathbb{E}(B_{\tau}^2).$$

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Thank you for your answer $\endgroup$ – Keen-ameteur Jun 24 '18 at 9:03
  • $\begingroup$ @Keen-ameteur You are welcome. $\endgroup$ – saz Jun 24 '18 at 11:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.