There is an urn with balls of one color. I draw one ball from this urn and put it back in. Somebody comes in and draws $m_0$ balls without replacement, when he is done, he then puts the balls he drew back into the urn, then somebody else comes in and draws $m_1$ times. Drawing a ball that nobody else drew gives me a value of x, if I draw the same ball as somebody else then we all have same chance of getting that object based on chance but only one of us will get it.

Let $C_i$ be the event that I draw the same ball as player i
Let a be the number of balls.

My expected value from this game is then: \begin{align} E = P(C_0 \cap C_1) \frac{1}{3}x + P(C_0 \cap \overline{C_1})\frac{1}{2}x+P(\overline{C_0} \cap C_1)\frac{1}{2}x + P(\overline{C_0} \cap \overline{C_1})x \\ =x (\frac{m_0}{a}\frac{m_1}{a} \frac{1}{3} + \frac{m_0}{a}(1-\frac{m_1}{a})\frac{1}{2} + (1-\frac{m_0}{a})\frac{m_1}{a}\frac{1}{2} + (1-\frac{m_0}{a})(1-\frac{m_1}{a}) ) \\ =x (\frac{m_0 m_1}{a^2} \frac{1}{3} + \frac{m_0-m_0 m_1}{a^2} \frac{1}{2} + \frac{m_1-m_0 m_1}{a^2} \frac{1}{2} + 1 -\frac{m_0}{a}-\frac{m_1}{a} + \frac{m_0m_1}{a} ) \\ =x (\frac{m_0 m_1}{a^2} \frac{1}{3} + \frac{m_0+m_1 }{a^2} +1 - \frac{m_1+m_0 }{a} ) \\ =x (\frac{m_0 m_1}{a^2} \frac{1}{3} + \frac{m_0+m_1 }{a^2}(1-a) +1 ) \end{align}

Question: How do I generalize the expected value for n players with each player having a specific number of draws? Trivially, if there are n players, the number of possibilities is $2^{n}$.

If there is also a better way to frame the problem than an urn problem, I am also interested.

Edit: Maybe a more useful way of phrasing probabilities is: Let $t_0$ be the event that no other person draws the same ball and $t_k$ the event that k other players draw the ball.
So the expected value is now:

$$E = \sum_{j=0}^n P(t_j)\frac{x}{j+1}$$

edit2: For the first 2 probabilities:

$$P(t_0) = \prod_{i=0}^n(1-\frac{m_i}{a}) $$

$$P(t_1) = \sum_{j=0}^{n} \left[\prod_{i\neq j}^n(1-\frac{m_i}{a})\frac{m_j}{a} \right]=\sum^n_{j=0} \prod^{j-1}_{i=0} \left(1-\frac{m_i}{a} \right)\prod^n_{k=j+1} \left(1-\frac{m_k}{a} \right)\frac{m_{j}}{a} $$


Ok so I think I have the general formulation! Would appreciate if somebody can see if this is correct or find a better way to express this.

$$P(t_2) = \sum^{n-1}_{j_1 = 0} \left[\sum^n_{j_2= j_1+1}\prod^n_{i \neq j_1 \neq j_2 }\left(1-\frac{m_i}{a}\right)\frac{m_{j_1}}{a} \times \frac{m_{j_1+1}}{a} \right] $$

$$P(t_k) = \sum^{n-k+1}_{j = 0}\sum^{n-k+2}_{j +1}...\sum^{n-1}_{j +k-2}\sum^{n}_{j +k-1}\left[\prod^n_{i \neq j \neq j+1...\neq j +k-1 }\left(1-\frac{m_i}{a}\right) \right] \prod^{j+k-1}_{q = j}\left(\frac{m_q}{a} \right) $$

  • $\begingroup$ I think the number of other players in this formulation is n+1 here instead of n, need to check... $\endgroup$ – Dio May 19 '17 at 15:02

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.