Yale over Baylor. Little Rock over Purdue. Wichita State over Arizona. Gonzaga over Seton Hall. After the first day of NCAA tournament play, March is already turning mad with upsets. But while teams are busting brackets, professors in the Furman math department are studying them.
More specifically, Liz Bouzarth, John Harris and Kevin Hutson, all Ph.D.s, are interested in how sports leagues rank players and team, which has a major effect on which matchups are drawn up for the NCAA men’s basketball tournament.
The group does not, however, keep its rankings to itself. All three consult Peter Keating and Jordan Brenner of ESPN for their column “Giant-Killers,” which attempts to predict and justify NCAA tournament upsets. Hutson and Harris also rank Furman in football each year, submitting FCS rankings to the NCAA before the season begins, and have worked with MLB statistics to forecast a batter’s performance against a given pitcher.
The professors created opportunities to teach some of the math behind ranking and brackets to students. Hutson has taught a first-year seminar focused on the mathematics of games and gambling, and Harris has offered a May X course in ranking systems.
This year, Bouzarth and Hutson have alternated leading sections of MTH-160: Vectors and Matrices, in which they pushed students to apply class knowledge and some outside concepts of probability to make their own predictions and seatings.
As far as the NCAA tournament goes, members of Bouzarth’s current MTH-160 section drafted their brackets around spring break. Performance within their brackets will determine a small quiz grade.
But how well students do in making their brackets is not always in their hands. Hutson says it is hard to point to any outstanding bracket-maker because so much of it is deals with luck.
“This is kind of like playing the stock market,” he said. “Some years you are up, some years you are down.”
Statisticians and forecasters, though, have a few tools to help analyze matchups. Past performance is a big indicator, as well as strength of schedule and road-win percentage. But these things can never be used to make a definitive prediction.
“That’s why, in the work we do for ESPN, we give our resulting in terms of probabilities rather than concrete statements,” Hutson explained.
One tool Hutson and his partners use to factor in chance is the random number generator. For example, if he determines a team has a 27 percent chance of winning, Hutson would set bounds for the generator at 0 and 100 and produce a random number in-between. If the number chosen falls below 27, he would choose the team to win. If not, the other team, clearly the favorite, would be selected.
Though people claim to be “bracket experts,” the teachers all agree that there is no simple way to foresee the “Cinderella story” team of a given year. Hutson puts it simply: “We’re all just rolling dice.”