#11
|
|||
|
|||
Re: Probability of pocket pairs in 10-handed game
[ QUOTE ]
If you scroll all the way down to the bottom of the page , it tells you the probability that a larger pair is out given that you hold pocket pair X . His numbers are for n=10 so if you hold pocket deuces , then the probability someone holds a higher pocket pair is =~ 0.3269 + 0.0818 +0.01186=~ 0.42056 [/ QUOTE ] I'll be the nit here and say that the table on that page of Professor Alspach's site is incomplete (it only has columns for "one rank", "two ranks" and "three ranks" although it's true that the contribution for higher ranks will be small). Also, the bottom entries are 0.3269, 0.08186 and 0.01186 and the sum of these is 0.42062, which only represnts a lower bound. I'll post a calculation using the inclusion-exclusion method, but I'll probably be a bit lazy and round off to the nearest 10^(-10), instead of expressing the answer as an exact fraction. |
#12
|
|||
|
|||
Re: Probability of pocket pairs in 10-handed game
Multiplying (3/51) = 1/17 by 10 is just the start. You
have to also consider when two or more players have a pocket pair. There is a useful idea called the "inclusion-exclusion" principle. Suppose the n events A[1], A[2], ..., A[n] are given and let P(S) denote the probability of S, S*T denote the intersection of sets S and T, S+T dentoe the union of the sets S and T, and sigma(conditions; expression) denote summation notation Then, P(S[1]+S[2]+...+S[n]) = sigma(1<=j<=n; P(S[j]) ) -sigma(1<=j<k<=n; P(S[j]*S[k]) ) +sigma(1<=j<k<l<=n; P(S[j]*S[k]*S[l]) ) -... +[(-1)^(n-1)]P(S[1]*S[2]*...*S[n]) You could also find this on various websites, perhaps if this seems unclear. The idea is you simply subtract off all "double counting" errors (you double counted when two players had a pocket pair) but then add back "triple counting" adjustments, etc. For the problem in question, the probability that the first player gets a pocket pair is the same as that for each of the other players, if there are no other conditions (if we don't know anything about what other players hold). Similarly, for any two different sets of n players, the probability that each set of players all have pocket pairs is identical (doesn't matter which players are selected) where 2<=n<=10. Let A[j] be the event that the jth player has a pocket pair and let C(m,n) be the number of combinations of m objects taken n at a time (which is simply m!/[(n!)(m-n)!] ). Note that there are C(10,n) ways of choosing n players at a table of ten players. Thus, the probability that at least one player has a pocket pair in ten-handed hold'em is: 10*P(A[1]) -C(10,2)*P(A[1]*A[2]) +C(10,3)*P(A[1]*A[2]*A[3]) -C(10,4)*P(A[1]*A[2]*A[3]*A[4]) +C(10,5)*P(A[1]*A[2]*A[3]*A[4]*A[5]) -C(10,6)*P(A[1]*A[2]*A[3]*A[4]*A[5]*A[6]) +C(10,7)*P(A[1]*A[2]*A[3]*A[4]*A[5]*A[6]*A[7]) -C(10,8)*P(A[1]*A[2]*A[3]*A[4]*A[5]*A[6]*A[7]*A[8]) +C(10,9)*P(A[1]*A[2]*A[3]*A[4]*A[5]*A[6]*A[7]*A[8]*A[9]) -P(A[1]*A[2]*A[3]*A[4]*A[5]*A[6]*A[7]*A[8]*A[9]*A[10]) The first term is just (1/17)10 = 0.588235294118 The second term (negative) is: -45(1/17)((12*6+1)/C(50,2)) = (-45/17)(73/1225) = 657/4165 = -0.157743097239. The first two terms sum to 0.430492196879 The third term is: 120(1/17)[(1/1225)(72/1128)+(72/1225)(68/1128)] =4968/195755 = 0.025378662103 The first three terms sum to 0.455870858982 The fourth term (negative) is: -210(1/17)[(1/1225)(72/1128)(67/1035)+ (72/1225){(2/1128)(67/1035)+(66/1128)(63/1035)} =-6590808/2431277100 = -8718/3215975 =-0.002710841969 The sum of the first four terms is 0.453160017013 The fifth term is: 252(1/17)[(1/1225)(72/1128){(1/1035)(66/946)+ (66/1035)(62/946)} +(72/1225){(2/1128)*[(1/1035)(66/946)+(66/1035)(62/946)] +(66/1128)*[(3/1035)(62/946)+(60/1035)(58/946)]} =4616377920/(17*1225*1128*1035*946) =0.000200713119 The sum of the first five terms is 0.453360730132 The sixth term (negative) is: -210(1/17)[(1/1225)(72/1128){(1/1035)(66/946)(61/861) +(66/1035){(2/946)(61/861)+(60/946)(57/861)} } +(72/1225){(2/1128)*[(1/1035)(66/946)(61/861) +(66/1035){(2/946)(61/861)+(60/946)(57/861)}] +(66/1128)*[(3/1035){(2/946)(61/861)+(60/946)(57/861)} +(60/1035){(4/946)(57/861)+(54/946)(53/861)}]}] =-206075469600/[17*1225*1128*1035*946*861] = -0.000010406329 The sum of the first six terms is 0.453350323893 From the Bonferroni inequalities, the probability is strictly between the last two sums, and since the seventh term will be small (anyone dare compute this?), the probability is approximately 0.45335. Anyone care to check this? |
#13
|
|||
|
|||
Re: Probability of pocket pairs in 10-handed game
Hey Pooch , you must have had fun working out the sixth term :P
Everything looks right which is consistent with the formula I've provided . To be honest with you , I don't really like the way Brian computed the probability of multiple pocket pairs because all of that work isn't even necessary and it's much better to use the inclusion/exclusion like you did . I haven't checked all the little details , but did you compute the probability that only distinct ranks of pairs are out ? |
#14
|
|||
|
|||
Re: Probability of pocket pairs in 10-handed game
In my calculations, if you check carefully, they include any
identical pairs. Well, I thought I could continue, but I didn't think there would be much value to know this to more than five decimal places! |
|
|