[absinthe]

Hi Mike,

You can formulate this as a stochastic dynamic programming problem (google may help here, but some algorithms background will come in handy), where you're performing a minimax optimization (i.e. minimizing the maximum amount of damage your opponents would be able to do to you, given your actions). This could be coded fairly straightforwardly, though it would be a bit slow. Hold'em is complicated enough that you probably won't find a summary that establishes rules for every situation. Furthermore, this model assumes your opponent always makes the best decision, which usually isn't the case (it would be a breakeven game if it were, and a losing game if you model a rake).

For more than 2 players, a minimax optimization may not be right, as it pretty much codes a collusion between your opponents (i.e. one might bet and the other raise in order to max charge a draw, even though the raisor knows he's behind). This makes the problem more complicated still.

Hope this helps. If you need a reference for dynamic programming, I can probably find one, but all it really does is give you a formula that you'd have to implement to understand.