Let $W_t$ be standard Brownian motion. It is well known that $W_t^2-t$ is a martingale. One way to show this is by applying Ito's lemma to calculate that $d(W_t^2-t)/dt = 2W_t dW_t$, which has no drift. Therefore $W_t^2-t$ is a martingale. I am a novice in stochastic process so I want to ask which theorems one use in this proof?


From Itô's lemma you have :

$Y_t=W_t^2 - t= \int_0^t W_sdW_s$

So what you have here is that $Y_t$ is a local martingale. To prove that it is indeed a martingale it suffices to show that :

$\forall t>0, E[\langle Y\rangle_t]<\infty$

as you can check in lemma 3 which is not too hard I think.

Best regards

  • $\begingroup$ Added \langle and \rangle (hope you don't mind) and upvoted (hope you don't mind either...). $\endgroup$ – Did Oct 15 '13 at 9:49
  • $\begingroup$ @ Did : It's ok as long as it is not a mathematical mistake in my post :-). Regards $\endgroup$ – TheBridge Oct 15 '13 at 10:00
  • $\begingroup$ why is $Y_t$ a local martingale? $\endgroup$ – mastro Jul 14 '15 at 17:05
  • $\begingroup$ @ mastro : Because : $\forall t>0, E[\langle Y\rangle_t]<\infty$ which is enough to show that $Y$ is a martingale (once an easy calculation is done). Best regards. $\endgroup$ – TheBridge Jul 14 '15 at 20:33

We can use Corollary 3.2.6 in page 33 of Oksendal's book, where the definition of $\mathcal{V}(0,T)$ is in page 25. So to prove $\int_0^tW_sdW_s$ is a martingale, we can simply verify that $E(\int^t_0W_s^2ds)<\infty$, which is true since this integral is $\frac{1}{2}t^2$.


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.