← All projects

No-Limit Hold’em Alpha-Beta Pruning

Aug – Dec 2024
PythonNumPyMonte Carlo
Git Repo

Highlights

  • Employed adversarial search with alpha–beta pruning to navigate simplified no-limit Hold’em decision trees, reducing computational complexity by 2+ orders of magnitude in early-game positions.
  • Reduced the game space to < 1e-6 of the original via Monte Carlo rollouts, hand bucketing, action limiting, and a probabilistic hand-range model (complete-information approximation).