Abstract

The Daily Fantasy Sports industry has recently skyrocketed. Led by powerhouses FanDuel and DraftKings, the industry is estimated to be worth over a billion dollars. The games themselves have been declared "skill games," meaning there is an underlying strategy behind Daily Fantasy Sports. This anomaly can be seen simply by that fact that only a few top players typically win most of the cash prizes. In this project we provide a deep exploration into the data science that lies behind choosing a team in Daily Fantasy Football. With team salary restrictions, Daily Fantasy Sports present an interesting optimization problem: users must make difficult decisions on which players to play on a given week. We investigate these decisions using analytics and machine learning to both model and analyze past data. The main goals of the investigation are to gain a better understanding of what statistics best predict player performance and find how we can produce the team to give contestants the best chance to profit. Through the investigation, we present our findings on the impact of past player performance, salary cost, and past opponent performance on a player's future performance. Finally, we develop models to choose teams based on these statistics and evaluate the models in a simulated Daily Fantasy Sports environment.

Advisor

Pasteur, R. Drew

Second Advisor

Visa, Sofia

Department

Computer Science; Mathematics

Disciplines

Artificial Intelligence and Robotics | Statistics and Probability

Publication Date

2017

Degree Granted

Bachelor of Arts

Document Type

Senior Independent Study Thesis

Share

COinS
 

© Copyright 2017 Conor R. Maley