Given prob space $(\Omega, \mathscr{F}, P)$ and a Wiener process $(W_t)_{t \geq 0}$, define filtration $\mathscr{F}_t = \sigma(W_u : u \leq t)$

Let 0 < p < q < r. Determine $E[W_p W_q W_r]$.

My attempt:

$0 = E[(W_r-W_q)(W_q-W_p)(W_p)]$

$\to E[W_p W_q W_r] = E[W_r W_p^2 + W_pW_q^2 + W_qW_p^2]$

$\to E[W_p W_q W_r] = E[(W_r+W_q) W_p^2 + W_pW_q^2]$

$\to E[W_p W_q W_r] = E[E[(W_r+W_q) W_p^2 + W_pW_q^2]|\mathscr{F_p}]$

$\to E[W_p W_q W_r] = E[W_p^2E[(W_r+W_q)|\mathscr{F_p}] + E[W_pE[(W_q^2)|\mathscr{F_p}]]$

$\to E[W_p W_q W_r] = ...0$ ?

It looks like the stuff are $\mathscr{F_p}$-measurable? $E[(W_r+W_q)]=0=E[W_p]$

I don't know. Help please? :(


3 Answers 3


I think you are on the right track here.

You made a sign error in the first line, unfortunately: $$E[W_p W_q W_r] = E[W_r W_p^2 + W_pW_q^2 - W_qW_p^2]=\\ E[(W_r-W_q)W_p^2]+E[W_pW_q^2]= E[W_pW_q^2] $$ The first term is $0$ by independence (as $p<\text{min}(r,q)$ and the square does not affect independence).

To take care of the second term we do the standard expansion trick: $$E[W_pW_q^2] = E[W_p(W_q-W_p)^2]+2E[W_qW_p^2]-E[W_p^3]$$

Now, the first term is $0$ again, by independence. For the third term we use that the third central moment of a normal distibution is$ 0$. The second term is also $0$, as a simple calculation shows: $$ E[W_qW_p^2] = E[(W_q - W_p)W_p^2]+E[W_p^3] = 0$$ again, by independence of increments.

So it turns out that $$E[W_pW_qW_r]=0.$$

Hope I got everything right. I must admit, I expected a different result.

  • $\begingroup$ We also have $E(W_t^3)=0$ which would aswell be consistent to $E(W_t^2)=t$ for $E(W_tW_p)=\min(t,p)=t,\,(t<p)$. $\endgroup$
    – emcor
    Oct 6, 2014 at 15:54
  • $\begingroup$ May you pls explain the third central moment more? Thanks so much vanguard2k $\endgroup$
    – BCLC
    Oct 6, 2014 at 19:06
  • 1
    $\begingroup$ For a normal distributed rv $X ~ N(\mu,\sigma)$, $E[(X-\mu)^3] = \mu^3+3\mu\sigma^2$ (just search for moment normal). In our case, $\mu=0$ and $\sigma = p$. Alternatively, you can calculate it by hand (several times integration by parts) or via the moment-generating function. $\endgroup$
    – vanguard2k
    Oct 7, 2014 at 6:42

We know that $(\tilde{W}_t) := (-W_t)$ is also a Wiener process so $$ E[W_pW_qW_r] = E[\tilde{W}_p\tilde{W}_q\tilde{W}_r] = (-1)^3E[W_pW_qW_r] $$ and that implies that $E[W_pW_qW_r] = 0$.

  • $\begingroup$ Ybl how do you know that they are equal? Thanks $\endgroup$
    – BCLC
    Oct 6, 2014 at 19:09
  • $\begingroup$ The product $W_pW_qW_r$ is not a Wiener process? $\endgroup$
    – emcor
    Oct 6, 2014 at 19:24
  • $\begingroup$ Emcor, are wiener processes equal to each other? Thanks $\endgroup$
    – BCLC
    Oct 7, 2014 at 2:22
  • 1
    $\begingroup$ @BCLC For the expectation they dont have to be equal (only identically distributed), YBL wants to show: $X=-X\Rightarrow X=0$. $\endgroup$
    – emcor
    Oct 7, 2014 at 12:11
  • $\begingroup$ YBL, @emcor How do you know that they are identically distributed? I am just wondering how you know X = - X in the first place i.e. $E[WpWqWr] = E[-Wp*-Wq*-Wr]$ $\endgroup$
    – BCLC
    Oct 9, 2014 at 13:37

Note that $\{W_t \mid t \geq 0\}$ is a martingale. Then, for $0<p<q<r$, \begin{align*} E(W_pW_qW_r) &= E\Big( E(W_pW_qW_r \mid \mathcal{F}_q)\Big)\\ &=E\Big(W_pW_q E(W_r \mid \mathcal{F}_q)\Big)\\ &=E\Big(W_pW_q^2\Big)\\ &=E\Big(W_p(W_q-W_p+W_p)^2\Big)\\ &=E\Big(W_p(W_q-W_p)^2+W_p^3+2W_p^2(W_q-W_p) \Big)\\ &=E(W_p)E\Big((W_q-W_p)^2\Big)+E(W_p^3)+2E\big(W_p^2\big)E(W_q-W_p)\\ &=0, \end{align*} by noting that \begin{align*} E(W_p) = E(W_p^3) = E(W_q-W_p) = 0. \end{align*}

  • $\begingroup$ Are Wp and Wq-Wp independent? $\endgroup$
    – BCLC
    Oct 10, 2014 at 10:40
  • 1
    $\begingroup$ For $p<q$, $W_p$ and $W_q-W_p$ are independent $\endgroup$
    – Gordon
    Oct 10, 2014 at 13:08

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.