Let $p$ be a probability in $]0,1[$, and let $(X^p_i)_{i \geq 1}$ be a i.i.d. family of variables with law $P(X=1)=p, P(X=-\frac{p}{1-p})=1-p$ (so that $E(X)=0$). Set $S^p_n=\sum_{k=1}^{n} X^p_k$ for $n\geq 1$, let $e(p,n)$ denote the probability that at least one of $S^p_1,S^p_2, \ldots ,S^p_n$ is positive (which means that the random walk reaches a positive value at least once during the first $n$ steps ). It is well known that for a fixed $p$, $e(p,n) \to 1$ when $n\to +\infty$.

Let $p\neq q$ be two probabilities. For large enough $n$, shall we have $e(n,p) \lt e(n,q)$ or $e(n,p) \gt e(n,q)$ ?

  • $\begingroup$ Are you comparing $p=0.01$ with $q=2/3$, or are you mainly interested in the case that both probabilities are on the same side of $1/2$? $\endgroup$ – Douglas Zare Aug 16 '11 at 1:32
  • $\begingroup$ @Douglas : I'm mainly interested in the case where both $p$ and $q$ are $<\frac{1}{2}$ (and $p$ and $q$ represent different gambling strategies). $\endgroup$ – Ewan Delanoy Aug 16 '11 at 5:12
  • $\begingroup$ If you restrict to the case where $1/p$ and $1/q$ are integers, then the analysis is nicer. Instead of saying some $S_i^p > 0$, the condition is that some $S_i^p >= 1$. Then $1$ has different sizes after you rescale by the standard deviations of the walks. $\endgroup$ – Douglas Zare Aug 17 '11 at 11:08
  • $\begingroup$ If $p$ and $q$ are both rational, the random walk is on $\frac{1}{n}.{\mathbb Z}$ for some $n$. $\endgroup$ – Ewan Delanoy Aug 17 '11 at 15:53

For the first question: yes, suitably scaled, the sequence $S^p_n$ tend to Brownian motion, which is positive infinitely often with probability 1. More precisely, for any fixed $p$, $e(n,p)$ will be asymptotically roughly $1-C_p n^{-\frac12}$.

As for the second question: it really depends on the range of $p$, $q$ and $n$. For fixed $n$ it is not monotone in $p$: for example, when $p$ is close to 1 we have $e(n,p)=1-p$, so it looks like it's decreasing, but when $p\approx \frac12$ we have a jump in $e(2,p)$ from around $\frac14$ to around $\frac12$.

The thing is that you have two contributing factors here: the probability of crossing 0 in the first few steps (how few depends on $p$) which is mostly a combinatorial question with incontinuity points, and the longer range behavior which is determined by the constant $C_p$ above. I guess that if you started the random walk at some distance from 0, you'd get mostly the latter factor.

  • $\begingroup$ The Central Limit Theorem tells us that the distribution of ${S_n}^p$ converges to $N(0,1)$. In what sense can it be said that tends to Brownian motion? Where does the $C_p$ constant come from? $\endgroup$ – Ewan Delanoy Aug 16 '11 at 5:09
  • $\begingroup$ I'm talking about $S^p_n$ as a sequence ($p$ fixed,$n$ goes from $1$ to $N$). When you rescale it you get a Wiener Process (en.wikipedia.org/wiki/Random_walk#Relation_to_Wiener_process). The constant comes from the variance of a single step of the random walk. $\endgroup$ – Ori Gurel-Gurevich Aug 16 '11 at 22:45

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.