I am currently engaging a research that would really use your help. I am considering add a brownian-type shock to a "fraction" $\theta \in [0,1]$, for example $$d\theta_{t} = \sigma \theta_{t} dB_{t}$$ where $\sigma$ is a constant.
However, the standard brownian motion does not necessary limit $\theta \in [0,1]$, I am just wondering if there is any stochastic process or transformation of Brownian motion truncation method that can make $\theta \in [0,1]$.

I think the simplest method is truncating directly, i.e. whenever the resulting $\theta>1$ (or smaller than $0$), we can make them truncated at $1$ or $0$, but I am pretty sure this process does not satisfy the standard properties of brownian motion (e.g. use Ito's lemma involving $\theta$), or does it?

Any help would be extremely appreciated!!

Thank you


You can very well choose a form $$ d\theta_t = f(\theta_t) dB_t $$ where $f(0) = f(1) = 0$. This way, when you approach 0 or 1, the speed decreases you you can never get out of $[0,1]$.

For a documented example, look for the Wright-Fisher diffusion.

  • $\begingroup$ Hi @mookid, Thank you for your reply, I am wondering if the following example works $$d\theta_{t} = \sigma \theta_{t} (1-\theta_{t})dB_{t}$$ Not 100% sure $\endgroup$ – EconGuy Oct 26 '14 at 22:51
  • $\begingroup$ It does!${}{}{}$ you can also have a drift part. If it is zero as well on the drift part, it remains in $[0,1]$. It may or may not be possible to arrive at the barrier in finite time. But if you attain the barrier you can't move any more. $\endgroup$ – mookid Oct 26 '14 at 22:51
  • $\begingroup$ It might be a stupid question but since theoretically $dB$ is not necessarily between 0 and 1, and if $dB = 10$, and $\theta = 0.5$, let $\sigma$ = 1, then we have $d\theta = 2.5$, I am just trying to think about discrete time analogue.. sorry it might sound stupid but I try to understand it, thank you! $\endgroup$ – EconGuy Oct 26 '14 at 23:00
  • $\begingroup$ remember that the trajectories of the Brownian motion are continuous! $\endgroup$ – mookid Oct 26 '14 at 23:03
  • $\begingroup$ You are welcome! $\endgroup$ – mookid Oct 26 '14 at 23:05

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.