Honors Theses

Creating a More Meaningful AI Tournament: Statistical Insights from the Game of Catan


Tam Nguyen, Ph.D.


Computer Science

Publication Date


Document Type

Honors Thesis


As the 14th most popular board game of all time, Catan distinguishes itself through a unique combination of randomness, strategy, and player interaction. This study explores the application of artificial intelligence (AI) to the game of Catan, utilizing the Catanatron framework to compare seven unique agents through an AI tournament. Central to this analysis is the application of statistically significant methods for comparison. Establishing quantifiable differences in the relative strength between agents is paramount in creating meaningful, repeatable findings in AI research. By generating statistically significant results, this study aims to create a foundation for comparing future agents created to play the game of Catan. Furthermore, the methods used for comparing agents in this study can be applied to similar games. This study adopts a tournament format to compare the agents, creating groups based on the initial findings of Bryan Collazo, creator of Catanatron. I placed Collazo’s original agents in opposition to each other. The first tournament round employing these agents builds a foundation for larger simulations. This study identifies the strongest agents, analyzes the nuances of their strategy, and quantifies their relative strength by looking at the statistical significance of the results. The results of this analysis contribute to AI research by producing meaningful comparisons between agents and providing a framework for future comparison that can extend to similar multiplayer games.

Permission Statement

This item is protected by copyright law (Title 17, U.S. Code) and may only be used for noncommercial, educational, and scholarly purposes.


Undergraduate research

This document is currently not available here.