Poker-AI.org

I am computing the optimal strategy for Leduc Holdem using the method in Fast Algorithms for Finding Randomized Strategies in Game Trees.

Leduc Holdem is played as follows:

• The deck consists of (J, J, Q, Q, K, K).
• Each player gets 1 card.
• There are two betting rounds, and the total number of raises in each round is at most 2.
• In the second round, one card is revealed on the table and this is used to create a hand.
• There are two types of hands: pair and highest card.
• There are three moves: call, raise, and fold.
• Each of the two players antes 1.
• In the first round, the betting amount is 2 (including the ante for the first bet). In the second round, it is 4.

This gives a total of 144 information states per player.

My result is that the first player expects to lose about 0.069555 per round. Has anyone analyzed this game and can confirm this? I have obviously tested my code on smaller games like Kuhn poker and gotten the correct results.