Boston is considered by many to be the center of the sports analytics universe, and many here on campus have played a substantial role in that. Last spring, Harvard College Sports Analytics Collective (HSAC) junior co-President John Ezekowitz was featured on the cover of the Boston Globe Magazine. Sophomore Andrew Mooney writes an aptly named column, “Stats Driven,” for Boston.com, and senior co-President David Roher, also a coxswain for Harvard lightweight crew, regularly contributes for Deadspin.
HSAC members publish independent research on our blog and are also engaged in sports analytics through the writing of academic papers, collaboration with other media outlets, and even employment with professional teams. Our presence in the sports analytics community at-large is continuously growing, but our presence on campus is minimal at best.
Part of this is due to HSAC’s lack of commentary on Crimson athletics. Here follows a short introduction to sports analytics’ applications to the Harvard fall season:
One of the simplest, and most commonly used, measures of a team’s quality–other than its straight-up win-loss record–is Bill James’ Pythagorean Win Expectation. The stat, more commonly referred to as a team’s “Pythag,” predicts a team’s winning percentage by dividing points scored squared over points scored squared plus points allowed squared. This is a more accurate estimation of a team’s quality because it values a 5-0 win greater than a 4-3 win and attempts to filter out “luck.” The football team, with 331 points scored and 144 allowed, has a PWE of .841 and a real win percentage of .875. The women’s soccer team has a PWE of .667 and a real win percentage of .735 with 34 goals and 24 points allowed. The field hockey team has scored 34 goals and allowed 37 for a PWE of .458 and a real win percentage of .471.
The validity of the application of Pythagorean win expectation to sports other than baseball is unclear. Houston Rockets General Manager Daryl Morey, a Northwestern and MIT grad, created a pythag expectation formula for basketball that multiplies points for and against to the power of 13.91 rather than two. But as a rough estimate of a team’s “luck,” the baseball model does seem to do a good job.
Another popular method is attempting to evaluate a player’s value to a team. This has resulted in statistics such as Wins Above Replacement Player (WAR) in baseball and Defense-adjusted Value Over Average (DVOA) in football.
Having covered field hockey this season, I can tell you anecdotally that there were numerous times when junior goalie Cynthia Tassopoulos single-handedly kept the Crimson in the game. Harvard gave up 69 more shots on goal to its opponents than it managed to take, while only allowing three more goals. That amounts to a disparity in save percentage of .806 for Tassopoulos and .754 for the Crimson’s opponents.
This is as much an indictment of the Harvard defense as it is a celebration of Tassopoulos’ performance. If you were to replace Tassopoulos with the “average” Ivy-League goalie, who would have a save percentage of .735, the Crimson would be expected to give up 55 goals. Using the Pythag formula, that means it would be expected to have a winning percentage of just .276. Over a 17-game season, that amounts to roughly a 5-12 record, meaning Tassopoulos was single-handedly worth an astonishing three of Harvard’s eight wins. That doesn’t even take into account her much-lauded role as an organizer of the defense.
Ivy-League co-champion Princeton had the lowest save percentage of any team in the Ancient Eight. If you were to replace Tigers goalie Christina Maida with Tassopoulos, 9-7 Princeton would be expected to go 13-3.
When we think of who the “Most Valuable Player” across a variety of sports is, we generally think of the quarterback on the winning football team or the leading scorer on the basketball team. Rarely do we think of the goalie on a losing team who gave up more goals than her team scored, but maybe we should.
— Staff writer Alexander Koenig can be reached at firstname.lastname@example.org.