1
$\begingroup$

A Gambler starts with $M\geq 0$ dollars in bank, starting at time $t=0$, he spends the $k^{\mathrm {th}}$ dollar at time $\sum_{i=1}^{k}A_i$ and he earns the $k^{\mathrm {th}}$ dollar at time $\sum_{i=1}^{k}J_i$, where independent sets of random variables $\{A_1, A_2, \ldots \}$ and $\{J_1, J_2,\ldots\}$ are samples of probability distribution function $f_A(x)$ and $f_J(x)$, respectively. Find $\mathbb{P}(L|A_1,\ldots,A_{L})$, the probability that the amount in his account becomes zero when he spends the $L^{\mathrm{th}}$ dollar, given the random variables $\{A_1, A_2, \ldots, A_{L} \}$.

I think since $L$ depends on $\{A_1, A_2, \ldots, A_{L} \}$ and $\{J_1, J_2, \ldots, J_{L-M} \}$, we first should find $\mathbb{P}(L|A_1,\ldots,A_{L}, J_1,\ldots,J_{L-M})$, then integrate it with respect to variables $J_1,\ldots,J_{L-M}$. Seems complicated.

$\endgroup$
  • $\begingroup$ So, what have you tried, show your progress.... $\endgroup$ – Brethlosze Oct 9 '17 at 6:22
1
$\begingroup$

Let us write $L=M+N$.If $ N<0$ the probability of depleting the account by spending $L$ dollars is zero. So let us consider the case $N\ge0$. Let us also write $$T_k =\sum_1^k A_i, \ \ S_k = \sum_1^k J_i.$$ Then if we think of how the process evolves in time, we need to always have a positive balance until time $T_L,$ and at that time the balance in the bank account should be $1$. This can be encoded in the following $N+1$ conditions: $$T_{M+1} > S_{1}, \ldots , T_{M+ N} \ge S_N, \text{ and } S_{N+1} \ge T_{M+ N}.$$ Now we can compute the conditional probability by integrating. As far as I can see though, there is no particularly nice closed for for this integral: $$\int_{-\infty}^{T_{M+1}} dx_1 \ldots \int_{-\infty}^{T_{M+N}} dx_N \int_{T_{M+N}}^{\infty} dx_{N+1} f_J (x_1) \cdots f_J(x_N - \sum^{N-1}_1 x_i) f_J(x_{N+1} - \sum^{N}_1x_i)$$

$\endgroup$

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.