Abstract
The use of AI and machine learning in sports is increasingly prevalent, including their use for in-game strategy and tactics. This paper reports on the use of machine learning techniques, applying it to analysis of U.S. Division I-A College Football overtime games. The present overtime rules for tie games in Division I-A college football was adopted in 1996. Previous research (Rosen and Wilson, 2007) found little to suggest that the predominantly used strategy of going on defense first was advantageous. Over the past decade, even with significant transformation of new offensive and defensive strategies, college football coaches still opt for the same conventional wisdom strategy. In revisiting this analysis of overtime games using both logistic regression and inductive learning/decision tree analysis, the study validates there remains no advantage to the defense first strategy in overtime. The study found evidence that point spread (as an indicator of team strength) and red zone offense performance of both teams were useful to predict game results. Additionally, by altering the decision-making “frame,” specific scenarios are illustrated where a coach can use these machine learning discovered relationships to influence end-of-regulation game decisions that may increase their likelihood of winning whether in regulation time or in overtime.
Highlights
The practical and widespread use of sports analytics continues to increase across the entire industry
We can see that the data does not support the conventional wisdom held by many that there is a significant advantage to being on defense first
The two machine learning approaches struggled to predict the outcome of overtime games, though the results were improved over pure chance
Summary
The practical and widespread use of sports analytics continues to increase across the entire industry. Artificial intelligence and machine learning techniques continue to be adopted in areas involving scouting and recruiting, training and performance analysis, revenue management strategies, and broadcasting and streaming decisions, in addition to strategic and tactical decisions “on the field” (Rein and Memmert, 2016; Joshi, 2019). This use of analytics transcends different sports, including those who have a long history of many statistics (such as baseball) to those which are more nascent in the use of big data (e.g., de Leeuw et al, 2018). Machine learning approaches (regression and decision tree algorithms) are applied to historical game data, resulting in additional insight into overtime game outcomes that could influence end-of-game decisions
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
