By Peter Rooney
In the basement of Amherst College’s Alumni Gymnasium, men’s soccer coach Justin Serpone is on a mission to introduce students to the analytical side of college athletics, and continue building a pipeline from Amherst College to the business side of sports.
“There’s been a lot of stories about baseball GMs like Ben Cherington ‘96, Neil Huntington ‘91 and Dan Duquette ’80 being from Amherst College,” said Serpone on a recent afternoon, his angular frame dressed in a black warm-up suit. “This is about filling in the gaps, getting students in at the ground floor and making opportunities happen.”
Serpone launched a fledgling program last fall, and has slowly been ramping it up with the support of Amherst LEADS, the college’s leadership development initiative. he recently held the inaugural Amherst Sports Analytics Forum, which featured presentations by guests like Smith College mathematics professor and former Mets statistician Ben Baumer, and several Amherst students discussed sports analytics projects that are in progress.
Among them was Megan Robertson ’15, a key member of the women’s basketball team and a statistics major. She’s working with Kevin Connors ’17 a track athlete and “huge basketball fan” to see if a series of four metrics developed by NBA analytics pioneer Dean Oliver have any relevance at the D-3 hoops level. She and Connors are wading through performance statistics for Final Four men’s and women’s teams for the past 10 years, with the aim of presenting an accessible, informative and hopefully valuable report to men’s coach David Hixon and women’s coach G. P. Gromacki.
“We may find that defense is more important, or that we need to shoot more free throws,” Robertson said. “I think this could be a huge contribution because I don’t think too many other teams are looking at statistics at this level in a way that can be useful for them.”
Also in Serpone’s office is Emily Horwitz ’17, who’s been spending several hours per week this semester scraping data from the last decade of field hockey performance statistics to see if she can uncover scoring trends.
“So far I’ve looked at goals scored versus our records, our goals scored every ten minutes to see if there’s a certain point where we tend to score or get scored against,” she said. “Also today I looked at what our winning percentage was when we scored first and when the other team scored first.”
“I found that 87 percent of the time if we score first we win.”
This gets Serpone excited, and he jumps out of his chair.
“What’s interesting is we did the same thing with our program!” he says. “We figured out that 95 percent of the time when we score first, we win. But 50 percent of the time when the other team scores first, we win.”
Based on those results, Serpone said he had a deep conversation with Jeremy Kesselhaut ’16, a mathematics major who has been analyzing men’s soccer team statistics since Serpone arrived in 2007.
“My first reaction was, ‘That means we should put all of our effort into scoring first,” Serpone said. “But Jeremy said, ‘Based on your record you’re going to win 83 percent of the time anyway, so don’t take the chance of trying to score first.’
“When I heard that, I thought, ‘Wow, that’s analytics in action,’” he said. “’You’re making coaches think more about what the numbers mean.’”
Kesselhaut said he enjoys panning for valuable nuggets of insight from the mountains of soccer data he’s been analyzing. Here’s another one of his findings: if the soccer team scores three or more goals in a game, there’s a 100 percent chance they’ll win the game.
"Although I may not be on a sport's team here at Amherst, I still care a whole lot about sports,” Kesselhaut said. “I am a big believer that a competitive edge can be obtained by sifting through data. You just have to look carefully."
There are no formal sports analytics courses yet at Amherst, though the college recently approved a new statistics major, and new statistics professor Nicholas Horton has been working with students interested in sports analytics, including Robertson.
“It’s really interesting stuff,” Horton said. “It’s connecting math, computer science, statistics and psychology to figure things out like whether to pay your goalie more, or to trade him.”
Serpone said he’s happy with the interest that students have shown in sports analytics. Now he’s thinking about how to expand the program to provide even more opportunities for students with a deep interest in the field.
As for how far analytics will guide the decision-making of coaches, Serpone said he’s open-minded, up to a point.
“I think this is important, and statistics can help you formulate thoughts and theories,” he said. “But it’s still players playing a game and everything is still based on one moment. We have to remember that, too.”