1) Obviously there was a big assumption in this calculation: no one folds. That is, of course, ridiculous. In fact, if your opponent has T3 you'll probably never have to beat his hand to actually win (whereas if he has AA, good luck getting him to fold). So the fact that you usually don't have to beat the hands at the bottom of the list to win probably changes the rankings. However, this seemed like a good spot to begin, and hopefully I can learn something from it.
2) As expected, suited connectors are clearly more valuable in a ten-handed game than a two-handed game.
3) Pairs are clearly more valuable in a two-handed game than a ten-handed game.
4) I may be playing too loose in my home game. I usually play a five-handed game. If I want to be in one pot on average per round of play (this seems reasonable to me), I should be playing cards only from the top 20% of the list. So I should probably consider not playing K9, A3s, or A8 anymore. Well, since position is important I should probably play only top 5% early position and maybe top 30-35% late position. Playing K9 in the small blind with a bunch of callers behind me is not looking so smart right now however...
5) 50 million hands may not have been enough to get the exact ordering in this question. Notice that a pair of 2s are actually ranked above a pair of 3s in the ten-handed game. If you refer to the percentages, they have almost exactly the same probability of winning (to all the decimals I calculated). I'm sure if I let the program run longer, this ordering would flip.
6) I'm curious to see the other games I've left out (three-handed game, four-handed game, etc). I'm going to turn the computer back on & see if I can't get lists for all games from two-handed to twelve-handed (this may take a while!)
7) I think my next simulation will be to fold weak hands and see how much the ordering changes in this list.