From: Thomas Barclay <Thomas.Barclay@s...>
Date: Mon, 24 Oct 2005 17:24:55 -0400
Subject: [GZG] The pain of dice stats
By exhaustive cases Yielding 19: D1: Nothing D2: 19 points D1: 1 points D2: 18 points D1: 2 points D2: 17 points D1: 3 points D2: 16 points etc. and then back down the other side D1: 9 points D2: 10 points D1: 10 points D2: 9 points D1: 11 points D2: 8 points etc. downto D1: 19 points D2: Nothing That would cover all possible combinations leading to 19 points. Then, it wouldn't be that hard to determine the possible ways to generate each of these scores per die. Usually, there is only one way. The sum of these chances vs. the sum of all possible outcomes on two dice will tell you how likely things are to happen. Using a little Excel wizardry D1 Rslt D2 Rslt Odds D1 Odds D2 Combined Odds D1 + D2 0 19 0.5 3.27456E-08 1.63728E-08 1 18 0.33 9.9229E-08 3.27456E-08 2 17 0.166666667 1.96473E-07 3.27456E-08 3 16 0.055 5.95374E-07 3.27456E-08 4 15 0.027777778 1.17884E-06 3.27456E-08 5 14 0.009166667 3.57225E-06 3.27456E-08 6 13 0.00462963 7.07305E-06 3.27456E-08 7 12 0.001527778 2.14335E-05 3.27456E-08 8 11 0.000771605 4.24383E-05 3.27456E-08 9 10 0.00025463 0.000128601 3.27456E-08 10 9 0.000128601 0.00025463 3.27456E-08 11 8 4.24383E-05 0.000771605 3.27456E-08 12 7 2.14335E-05 0.001527778 3.27456E-08 13 6 7.07305E-06 0.00462963 3.27456E-08 14 5 3.57225E-06 0.009166667 3.27456E-08 15 4 1.17884E-06 0.027777778 3.27456E-08 16 3 5.95374E-07 0.055 3.27456E-08 17 2 1.96473E-07 0.166666667 3.27456E-08 18 1 9.9229E-08 0.33 3.27456E-08 19 0 3.27456E-08 0.5 1.63728E-08 Total Results leading to 19 points: 6.22166E-07 In all of these cases, once you break the two dice apart, their is precisely one way to generate the point score with one die roll sequence. To get 1, I need to roll 4 or 5 (ergo 0.5). To get 2, I need to roll 6. To get 3, I need to roll 6 followed by 4 or 5, etc. Treating each dice separately, and multiplying the probabilities of a given outcome together should give us the total for that particular combination of points. Then sum these up for the total. I make this at 1 in 1607288.038, given some rounding errors. So call it 1 in 1.6 million in round figures. Pretty unlikely. Can we put this bad boy to rest now? Regardless of the math, it's a pretty freakin' unlikely circumstance and to make rules from it seems a poor choice. (YMMV). TomB