An Examination of Win Probability in NCAA Division I Football Overtime and Its Applications for Evaluating Coaching Decisions

Name: Matt Ulishney
Major: Mathematics
Minor: Statistical and Data Sciences
Advisors: Drew Pasteur, Qimin Huang

The increasing use of analytics in sports has changed the way organizations, coaches, and players approach game planning, strategies, and evaluation of performance. One common metric to describe a team’s chances of winning in progress contests is win probability, or WP. Win probability is comprised of many game state variables, including down, distance, yards to goal, and score differential, along with team strength ratings such as point spread and offensive or defensive ratings. In this study, we will examine what variables are useful in determining win probability for in progress NCAA Division I football games in the overtime periods. We use logistic regression and decision trees to create win probability models using various game state and team strength variables. Our results are promising, as they perform well in describing a team’s win probability given various overtime game state situations. We can use our results to evaluate coaching decisions made in the overtime periods, in addition to determining which plays had the greatest impact on a team’s chances of winning the game.

Posted in Comments Enabled, Independent Study, Symposium 2023 on April 14, 2023.

One response to “An Examination of Win Probability in NCAA Division I Football Overtime and Its Applications for Evaluating Coaching Decisions”

1. Mason Gilton says:

   April 21, 2023 at 4:26 pm

   I personally found the findings of this study very important and impactful to the game at large. As a connoisseur of the game of football, these data can contribute to the way coaches make decisions in the future and affect how aggressive teams chose to be in certain scenarios.