Centre research team runs the play on football probabilities
With an undefeated football season underway, Centre College is engaging students in the popular pastime both on and off the field. Associate Professor of Mathematics Jeffrey Heath and David Newton ’15 (pictured above) are studying the stochastic modeling and simulation of football. This project is a continuation of Heath’s summer “sports analytics” research, in which he and his students used mathematics and probabilities to study baseball and basketball. Heath and Newtown are now turning their attention to football.
“I wanted to get into football for a couple of reasons. One reason is that sports analytics in football is a bit more in its infancy than in baseball and basketball,” Heath explains. “Two, I think that the discrete nature of football is a fun area in which to work. And I love probabilities. That’s all it is, just one probability project.”
The pair uses 11 years worth of National Football League (NFL) data to model football probabilities in order to simulate game play. Through their simulations, they can run a game “thousands of times” to see which plays and decisions are optimal.
Their statistical analysis could be used to provide coaches with the information they need to make an “informed game-time decision” on which plays to run.
“Typically, when you’re in a game, it’s a crucial point to decide, ‘Do we punt on the fourth down or do we decide to run?’ There’s risk either way—you might want to play it safe or you might want to run for it,” Newton says. “If we’re deciding if we want to punt or go for it, we can simulate ten thousand games and see that we won a lot of games when we went for it and we lost a lot of games when we punted, so maybe we shouldn’t do that.”
Newton presented their research at the Bluegrass Undergraduate Mathematics Symposium, held Oct. 4 on Centre’s campus. Heath hopes the summer research experience provided Newton with “hands-on experience with real-world data.”
“Data analytics is a huge, growing field, and in a classroom setting, you typically aren’t exposed to it. You don’t have the time, and a lot of the students don’t have the background,” Heath says. “[Newton] has that now, and that’s an attractive skill. He doesn’t have to stay in sports, but he’s used to processing data, and he has some simulation background. There are a lot of skills that he’s picked up.”
Newton believes that the “collaborative” environment at Centre was conducive to completing this research.
“I’ve learned some knowledge here and there, whether it’s about football or different statistical techniques I didn’t know from class,” Newton says. “Since I have to code for several hours a day, that process has been very helpful just in my general problem-solving ability. We’ve written thousands and thousands of lines of code this summer. I feel like that has benefited me a lot. There are more tangible ways of learning how to extract data and make distributions. The practice from doing it so much has really paid off.”
Throughout the remainder of the football season, Heath hopes that they can “add more layers of analysis” to their simulations, specifically to better assess player performance.
“Ultimately, what we’re looking for in an offensive player is if that player increases the team’s expected point total on a drive. How can their team score more, or are you preventing them from doing that?” Heath says. “We’re finding ways to isolate this information. We haven’t done much of that, because we spent so much time building the model. We’re just now scratching the surface.”
by Hayley Hoffman ’16