Instead of evaluating a sample of games (e.g. the several 1000 a particular player may have come across), I'm trying to play "all" possible games to create a "perfect" strategy.
It's early days yet. I have a number of programs that look at various aspects of the basic game. The major components of the strategy are evaluating the hole cards and betting strategies for different tables. I'm more interested in tournament play than winning single games.
As a taste of what I'm doing, the following is a table of approximate "values" current algorithms recommend for each pair of hole cards.
Positive values indicate a better-than-even prospect of a win, assuming the showdown will be a head to head against one other player. Negative values indicate folding is probably in order if we're playing an early tournament round. Higher values indicate more aggressive betting is likely worthwhile.
Note some of the non-obvious recommendations.
As my software improves and/or accumulated cpu time increases, results are likely to change.
If you have questions or random constructive comments, feel free to drop me a line.
