Mahjong enjoys its status as the national game of China. The way it is played is philosophically described as: to create order out of chaos based on random drawings of tiles.

This study focuses on one specific type of mahjong: Sichuan mahjong. Sichuan mahjong is one of the most famous mahjongs in the world, and its unique rules lead to the two-player mahjong situation, which remains a gap in scientific research.

This report presents a Markov chain model for Sichuan mahjong, focusing on the quantitative measure of the game state. We took the combinatorial theory and algorithmic approaches to understand the game states. Based on the game state measure, we calculated the winning probability and expected number of game rounds in the two-player situation, and compared them with the experiment results. The results demonstrated the playability of Sichuan mahjong.

We also investigated the difference between aggressive and conservative players, and simulated the aggressive player’s strategy in the calculation of winning probability and expected number of rounds.

The quantitative methods for game state measure contribute to applications in future mahjong AI, which provides the players with a broader understanding of Sichuan mahjong.