2
$\begingroup$

Let $(B_t)_{t > 0}$ be a Brownian motion and let $M_t = \sup_{0 \leq s \leq t} B_s$. I was wondering whether $M_t - |B_t|$ was a Brownian motion as well ? If so, why ?

From this question, I would naively expect $M_t - |B_t|$ to behave like $B_t$.

$\endgroup$

1 Answer 1

4
$\begingroup$

For $x>0$, we have the disjoint union $$\{B_t<-x\}=\{M_t-|B_t|<-x\} \cup \{-x> B_t \ge-x-M_t\}$$ so $$P[B_t<-x]>P[M_t-|B_t|<-x] \,.$$

Addendum: Note that $$\{-x> B_t \ge-x-M_t\} \supset \{2>B_{t/2}>1\}\cap \{-x>B_{t}>-x-1\}$$ and using the Markov property and the monotonicity of the Gaussian density, $$P[ 2>B_{t/2}>1 \; \text{and} \; -x>B_{t}>-x-1]$$ $$ \ge P[2>B_{t/2}>1] \cdot \inf_{h\in [1,2]} P[-h-x>B_{t/2}>-h-1-x] $$ $$ = P[2>B_{t/2}>1] \cdot P[-2-x>B_{t/2}>-3-x] >0\,.$$

$\endgroup$
2
  • $\begingroup$ Thank you ! I'm just not sure on how to justify properly that the second set in the union has a nonzero probability. Would you mind clarifying this last step ? $\endgroup$
    – Desura
    Commented Jun 25, 2022 at 7:30
  • $\begingroup$ @Desura Is it clear now? I added an explanation. $\endgroup$ Commented Jun 25, 2022 at 7:59

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .