LEGITIMATE OR UNFAIR?: AN EVALUATION AND IMPROVEMENT OF THE COLLEGE FOOTBALL PLAYOFF USING LOGISTIC REGRESSION AND ADJACENCY MATRICES

Abstract

The world of college football has the unique challenge of picking the best teams in the football championship subdivision. Always up for debate, many fans, writers, and scholars have questioned whether the best teams in college football by season’s end are truly the best teams. This research explores the history of finding the best college football teams since the beginning of the NCAA, including the current College Football Playoff. Because of the current method’s subjectivity, a method consisting of four logistic regression models and a series of weights is used in an adjacency matrix to determine the best teams in the nation. The logistic regression models are based on game data from the top 25 teams during the first six seasons of play under the current College Football Playoff. With an average difference of 7.49 places between teams in both the College Football Playoff rankings and our rankings, our method only serves as a foundation for what could be a more objective way of determining which teams deserve to be at the top, particularly with logistic regression and more expansive game data.