For Lebesgue integrals one has the triangle inequality saying that for continuous functions let's say

$$\left\vert\int_0^t f(s) \ ds\right\vert \le \Vert f \Vert_{\infty} \int_0^t \ ds$$

Now if we consider an Ito integral, then

$$\left\vert\int_0^t f(s) \ dW(s)\right\vert \le \Vert f \Vert_{\infty} \vert \int_0^t \ dW(s)\vert$$

does not hold pointwise, but I was wondering whether this one holds probabilitically, i.e.

does there exist a constant $c(t)>0$ such that for all deterministic continuous $f$

$$\mathbb P\left(\vert W(t) \vert \Vert f \Vert_{\infty}\ge a\right)\ge c(t)\mathbb P\left(\left\vert\int_0^t f(s) \ dW(s)\right\vert \ge a\right)?$$


If the function $f$ is indeed deterministic, with $M:=\|f\|_\infty$ and $\sigma^2:=\int_0^t f(s)^2\,ds$, then $X:=\int_0^t f(s)\,dW(s)\sim N(0,\sigma^2)$, whereas $Y:=MW(t)\sim N(0,M^2t)$, and $k^2:=\sigma^2/(M^2t)\le1$. So, $X$ equals $kY$ in distribution, and so, $P(|Y|\ge a)\ge P(|X|\ge a)$ for all real $a$; that is, your inequality holds with $c(t)=1$.


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.