Monte Carlo Tree Search

Monte Carlo Tree Search (MCTS): A heuristic search algorithm that uses statistical sampling and random playouts to efficiently navigate massive decision spaces, most famously powering AlphaGo.

MCTS is a powerful heuristic search algorithm designed to conquer decision spaces too vast for exhaustive search: think $10^{170}$ possible states in the game of Go. It operates on a four-step iterative loop: Selection, Expansion, Simulation (random playout), and Backpropagation. The core innovation is the Upper Confidence Bound for Trees (UCT) formula, which intelligently balances **exploitation** of moves with high win rates against **exploration** of less-visited branches. This approach was famously combined with deep neural networks to create DeepMind's AlphaGo, which defeated the world champion in 2016, and its successor, AlphaZero, for games like Chess and Shogi.

https://en.wikipedia.org/wiki/Monte_Carlo_tree_search