-1

In this game, we use a standard shuffled deck of $52$ cards. The suits of each card do not play any part.

Take a card from the deck one by one. If the card's rank is equal to the number of cards you have taken you win. You may take up to $10$ cards in a single game.

To clarify the rules here are some winning game examples;

  1. $9164$
  2. $735626$
  3. $48K25$
  4. $1$
  5. $92$
  6. $218927299$

What are the odds of winning a single game?

  • 2
    Please edit to include your efforts. Also for clarity. I guess Aces have rank $1$, right? Are face cards assigned any numerical rank ? – lulu Sep 28 '23 at 13:31
  • As a way to get started, I suggest computing the expected number of "winners" in the ten draws, that's easy. Slightly harder (but still pretty easy) is to compute the variance of the number of winners. – lulu Sep 28 '23 at 13:34

1 Answers1

3

To make things easier, always draw 10 cards regardless if we win or not. We use inclusion-exclusion principle over the events $X_1,X_2,\dots X_{10}$ where $X_i$ is the event that position $i$ has a card of rank $i$.

We are interested in $\Pr(X_1\cup X_2\cup \dots X_{10})$ which expands as:

$$\Pr(X_1)+\Pr(X_2)+\dots+\Pr(X_{10})-\Pr(X_1\cap X_2)-\dots -\Pr(X_9\cap X_{10})+\Pr(X_1\cap X_2\cap X_3)+\dots \pm \Pr(X_1\cap X_2\cap \dots \cap X_{10})$$

Assuming $10$ is the only rank of card said to have value of $10$ and Jacks, Queens, and Kings do not count as working... massive simplifications can be made in the above thanks to symmetry.

$$10\Pr(X_1)-\binom{10}{2}\Pr(X_1\cap X_2) +\binom{10}{3}\Pr(X_1\cap X_2\cap X_3)\pm\dots\pm \binom{10}{k}\Pr(X_1\cap \dots \cap X_k)\pm\dots$$

The probability of $\Pr(X_1\cap X_2\cap \dots\cap X_k)$ will be that an ace is in the first spot, a 2 is in the second spot, on up until a $k$ is in the $k$'th spot. This expands by multiplication principle as $\frac{4}{52}\times \frac{4}{51}\times \cdots \times \frac{4}{52-k+1}$, noting that in the above we only care about hitting or not hitting the desired cards in the specified slots and any additional slots are ignored and can have any result including additional matches, making calculations easy.

We get then a probability of winning as:

$$\sum\limits_{k=1}^{10}(-1)^{k+1}\binom{10}{k}\frac{4^k}{52\frac{k}{~}}\approx 0.5481536$$ wolfram calculations

JMoravitz
  • 79,518
  • Thanks for the answers, I think this follow simpler to my first guess at an answer but I wondered if the past dealings had an impact? For example, the first spot is easy, 4/52, but the second spot isn't 4/51 if the first was a 2, it would be 3/51, right? Or does that not matter? – Ash Burlaczenko Sep 28 '23 at 14:02
  • I have a question regarding your simplifications in the spirit of $Pr(X_1)=Pr(X_2)=...=Pr(X_{10})$ and so on. I agree that $Pr(X_1)=\frac{4}{52}$ and if the first card is no 2, then also $Pr(X_2)=\frac{4}{51}$ but does your simplification take into account that if the first card is a 2 the probability changes $Pr(X_2)=\frac{3}{51}$? – Keine_Maschine Sep 28 '23 at 14:04
  • $\Pr(X_2)=\Pr(X_1)$ here. As alluded to, there is heavy amount of symmetry at play here. In some sense, yes past draws have an impact... but only if we bothered to keep track of what those past draws were and if we conditioned the probability on those earlier results. Without conditioning, they have no impact. The probability the second card is a $2$ is $\frac{4}{52}$. The probability the $k$'th card is a $k$ is $\frac{4}{52}$ for any $k\in {1,2,\dots,10}$ – JMoravitz Sep 28 '23 at 14:05
  • See https://math.stackexchange.com/questions/1287393/if-you-draw-two-cards-what-is-the-probability-that-the-second-card-is-a-queen – JMoravitz Sep 28 '23 at 14:06
  • With regards to why $\Pr(X_3\cap X_5\cap X_9) = \Pr(X_1\cap X_2\cap X_3)$ and similar... we can make the observation that we can draw the cards out of order and it not impact the probabilities. We draw the third card from the deck, and then the fifth, and then the ninth, and then go back and fill out the first, second, fourth, etc... This does not affect the probabilities. You can prove this by finding a bijection between the scenarios if you insist. Thanks to the simplification, it is clear why the probability is then $\frac{4}{52}\times \frac{4}{51}\times \frac{4}{50}$ – JMoravitz Sep 28 '23 at 14:12