Author

Kyle Saunders

Date of Award

2014

Document Type

Thesis

Degree Name

Bachelors

Department

Natural Sciences

First Advisor

McDonald, Patrick

Keywords

Football, Ranking Systems, Bowl Championship Series

Area of Concentration

Applied Mathematics

Abstract

Ranking college football teams is a difficult task, due to the large number of teams and relatively short schedule. Many ranking systems exist, but few publish their methodology. Even fewer contain any evaluation of the predictive accuracy of their rankings. This thesis develops three regression-based ranking systems and tests them for predictive accuracy against the six computer components of the Bowl Championship Series rankings, which were used to decide college football's national champion from 1998-2013. Two of the methods perform favorably compared to the BCS computer rankings, beating or tying all six. These relatively strong results, along with the scarcity of other such analyses within the literature, indicate that further research in this area could be fruitful.

Share

COinS