George Baldini, Lina McDermott
Faculty Sponsor: Dr. Raghu Ramanujan
We implement an agent to play Leduc Hold’em, an incomplete information poker variant. The agent employs counterfactual regret minimization in order to determine its next move. We measure our agent’s performance in terms of how much regret it has after each round of the game and its overall utility. Ultimately, we show that our agent is able to successfully play Leduc Hold’em.
