Poker-AI.org

Hello.

I'm looking at the "Opponent Modeling in Poker" paper (http://www.cs.virginia.edu/~evans/poker ... llings.pdf). On page 4 there is code for ppot and npot as well as an example breakdown for AdQc+3h4cJh.



Rows correspond to where the hand was on flop and columns correspond to where it went on river. My problem is that the paper says ppot_2 = 0.208. But when I put the numbers into the formula I get a slightly different number:
(91981+1036*0.5)/(8910+439560) = 0.20625...

Am I doing something wrong?

The only explaination I have is that the authors actually mean to have only 1/2 of the total ties in the denominator.

There is another article that has exactly such formula.

And here is another person that uses the formula with 1/2 ties in the denominator.

Did you try to play preflop with hand strenght ?
Is it better than a static approach like "Sklansky tables" or is it similar?

No, haven't tried that. So far have only been using E[HS] and E[HS^2] for preflop bucketing.

Why not just use an unabstracted preflop, assuming disconnected rounds?