With the basic framework to handle a game done, I’m starting to build intelligence into my Texas Hold’em AI. One of the first steps I’ve taken in this direction is to precalculate the preflop expectation of any of the 169 possible two card hands that a player can have in holdem. This is the chance of beating any particular player, one on one, if the game goes to the river. With this number it is trivial to calculate the preflop odds of beating N players, given a specific hand. These odds can then be used to decide whether to bet X dollars or fold, though I’ven’t decided on a particular algorithm for that yet.
Rather than sum over all possible outcomes, my preflop calculator uses a Monte Carlo approximation which works great for large sample size (as you can see in the screen shot).
