Optimal Multiplayer Yahtzee

A Nearly Optimal Computer Player in Multi-player Yahtzee
Jakub Pawlewicz
Institute of Informatics, Warsaw University
2010

Abstract

This study introduces a pioneering analysis of multi-player Yahtzee, a widely popular dice game, addressing the complexities of developing optimal computer strategies. While single-player strategies for maximizing average scores and specific score achievements are well-documented, multi-player strategies remain computationally intensive. Our research presents optimized implementations for single-player Yahtzee, significantly reducing calculation times. For the two-player game, we precisely estimate the resources needed to develop an optimal strategy and propose a heuristic approach for scenarios involving more players. Experimental results demonstrate that the heuristic strategy performs comparably to the optimal strategy in practice. Additionally, our analysis reveals that Yahtzee is largely a game of chance, with minimal advantage gained from optimal strategies against human opponents. This work advances the understanding of computational strategies in probabilistic games and provides a foundation for future research in multi-player game optimization.

More Resources

Game and Variance

An Optimal Strategy
for Yahtzee

Markov Variance and
Yahtzee Probability