1
$\begingroup$

I am trying to calculate the probability of taking from a regular deck of cards (52) 5 cards with 4 different suits, my go:

first we think about picking 'diamond' (just to make it more simple) it is: $\frac{13}{52}$
second we think about picking 'heart': $\frac{13}{51}$
same thing for the two other suits: $\frac{13}{50}$ and $\frac{13}{49}$ respectively.
The fifth card needs to be the same one of the suits, so: $\frac{12}{48}$ because we took each card from each suit, so we have only $12$ remaining in that same suit.

But we have to consider the order that we draw the cards, so we multiply by $5!$

And we have my final answer of: $5! \cdot \frac{13}{52} \cdot \frac{13}{51} \cdot \frac{13}{50} \cdot \frac{13}{49} \cdot \frac{12}{48}$

Sadly, this answer is only half the real answer: $\approx 0.26375$ mine gives: $\approx 0.13187$

Where does the $ \cdot 2$ need to come from? I don't see another orientation other than the suit we pick the second card to be of, so we don't need to multiply by $2$ but by $4C2 = 6$ and that gives me an answer that times $3$ larger than the real one, so in that case, where does the $\frac{1}{3}$ need to come from?

Thank you!

$\endgroup$
1
  • $\begingroup$ You have two mistakes which combine to give you half the right answer. First, in your last step, any suit will work if you have $4$ different suits in the first $4$ draws -- so $12/48$ makes your answer $1/4$ of what it "should be". Second, the method of choosing one card of each suit and then an arbitrary fifth card over-counts by a factor of two, because any hand, say 3S 3H 5D 7C 9C, gets counted twice, once with 7C in the first group of $4$ and once with 9C in the first group. $2*(1/4) = 1/2$ $\endgroup$
    – Ned
    May 12 '20 at 14:36
1
$\begingroup$

You could have picked either of the cards of the same suit first, so you’re counting each admissible hand exactly twice.

$\endgroup$
6
  • $\begingroup$ So, because theoretically I could have picked the card that has the same suit first, and not last I need to multiply everything by 2? $\endgroup$
    – CSch of x
    May 12 '20 at 9:47
  • 1
    $\begingroup$ @OUR: No, the opposite: Because of that, you need to divide by $2$, since you double-counted. $\endgroup$
    – joriki
    May 12 '20 at 10:04
  • $\begingroup$ But then I get an answer that is smaller by a factor of 4 :( $\endgroup$
    – CSch of x
    May 12 '20 at 10:07
  • 2
    $\begingroup$ @OUR You should still multiply the result by the number of suits (see also my answer). $\endgroup$
    – user
    May 12 '20 at 10:09
  • 2
    $\begingroup$ @OUR I added the explanation where your way fails. $\endgroup$
    – user
    May 12 '20 at 10:20
1
$\begingroup$

Obviously, the "correct" hand should contain two cards of a suit and one card of each of the other 3 suits. There are $\binom41$ ways to choose the suit with the double, and $\binom{13}2$ ways to choose the double out of the suit. Therefore the overall number of "correct" hands is: $$ \binom41\binom{13}2\binom{13}1^3. $$ Dividing this by the overall number of hands: $\binom{52}5$ you obtain the correct probability.

You way fails because of two reasons: first you double-count the hands and second your miss the factor 4 because the "last" card may be of any of 4 suits.

$\endgroup$
1
  • $\begingroup$ I see now, thank you! $\endgroup$
    – CSch of x
    May 12 '20 at 10:23
0
$\begingroup$

I was thinking that the problem could be solved by using the hypergeometric multivariate distribution

P(x,y,z,t) = [ 13Cx 13Cy 13Cz 13Ct ] / 52C5

where x,y,z,t are the random variables associated to the four disjoint sets of the suites

as the only 4 possible configurations of the random variables compatible with the requested probability are the following

x y z t

2 1 1 1 1 2 1 1 1 1 2 1 1 1 1 2

we find:

P = P(2,1,1,1)+P(1,2,1,1)+P(1,1,2,1)+P(1,1,1,2) = [ 13C2 13C1 13C1 13C1 ] x 4 / 52C5 = 26%

$\endgroup$

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.