Graduation Year

2025

Date of Submission

12-2025

Document Type

Campus Only Senior Thesis

Degree Name

Bachelor of Arts

Department

Economics

Reader 1

Peter Kelly

Rights Information

2025 Daniel Hirose

Abstract

Understanding how people make choices when money is at risk is a long-standing challenge in economics. Poker offers a rare chance to watch these decisions play out in real time while the stakes, uncertainty, and pressure are all visible. In this project, I compare actual betting behavior in online No-Limit Hold’em to the frequencies recommended by Game Theory Optimal (GTO) solvers. The goal is to see whether real players show the kinds of risk-averse tendencies predicted by theories like Prospect Theory and bounded rationality. Using a dataset of more than 8,000 hands that I played and recorded myself, I focus on specific betting situations, primarily flop c-bets and turn barrels, and match each one to a GTO benchmark for that exact board texture. I then run a combination of proportion tests and regression models to measure how far human behavior strays from GTO, and whether those deviations change with features of the board or the size of the pot. Instead of being overly cautious, players in my sample generally bet more than what GTO recommends. Despite this over-aggression, human betting still moves in the same direction as GTO’s guidance. The one place where risk sensitivity seems to appear is when the pot gets larger; over-betting becomes smaller on the turn, hinting that higher financial stakes may encourage more cautious play. Overall, the findings shed light on how real decision-makers behave in complex, high-pressure environments. They suggest that people simplify strategy through heuristics, follow broad optimal patterns, but often mis-scale their actual behavior - insights that matter not only for poker, but also for anyone studying financial decision-making, behavioral biases, or human responses to risk.

Download

This thesis is restricted to the Claremont Colleges current faculty, students, and staff.

Share

COinS