The MIT Sloan Sports Analytics Conference is more of a sports business conference these days, but the MIT association conveys a conference that also values rigorous and high-quality research. The Research Paper competition keeps the SSAC connected to this heritage as it evolves. This year the competition returns with tracks for publications on Baseball, Basketball, Other Sports, and the Business of Sports.
The competition is in two phases: the first phase whittles down the field on the basis of research abstracts submitted, and the survivors are invited to submit full papers that are reviewed by the SSAC panel. The eight semifinalists present their papers in the Research Paper session on Friday, and three finalists get the chance to do it again on Saturday. The winning submission receives at least USD 10000 and a guest column on ESPN’s FiveThirtyEight website. You can view the semifinalists and their papers on the SSAC website.
Basketball and baseball are always represented among the finalists (three basketball papers and two baseball papers), but there are papers in mixed martial arts, sports sponsorship, and soccer (for the fourth year in a row). I am amused that I complained in an infamous podcast in 2013 about the lack of soccer publications at Sloan, and now we’re in a period of soccer analytics research of highly publishable quality.
A scan of the authors reveals some familiar names and institutions. Luke Bornn’s research group at Simon Fraser University appears in the final eight for what I believe is the fourth year in a row. Patrick Lucey’s current and former research groups are also present (third year in a row), as is Jonathan Jensen from UNC-Chapel Hill. All of the authors in the Research Paper competition are affiliated with research universities or are closely tied to them (Disney Research is closely tied to Carnegie Mellon University, for example). I write this to show that it is very difficult to perform the type of research that gets noticed by the SSAC panel without collaborations and institutional support.
The SSAC was late in getting the papers online this year, so I probably won’t be able to read all of them before the competition starts on Friday. I have read almost all of the abstracts though, and I am looking forward to seeing the following presentations:
- “A switching dynamic generalized linear model to detect abnormal performances in Major League Baseball“, which is an interesting application of switching dynamical models to infer outlier performances in baseball as a function of a player’s age
- “Data-driven ghosting using deep imitation learning“, an application of deep learning algorithms to player positional data in soccer to develop metrics that compare expected to actual defensive positioning
- “Why do marketing partnerships end?“, which uses survival analysis to identify economic and market factors that increase the risk of dissolution of sports sponsorships
- “Possession sketches: mapping NBA strategies“, which applies topic modeling techniques in a clever way to identify types of plays in a basketball game and organize game possessions accordingly (I saw this presentation at CASSIS last September)
The other papers in the group include an attempt to describe three-point shooting styles in basketball in terms of a “basis vector” of attributes in all shooting phases, formulate new strategies for managing starting and relief pitchers by visiting teams in baseball, data mine characteristics of mixed martial art fighters to predict match outcomes, and use basketball analytics to increase interest in STEM fields among young athletes.
The poster event features some soccer-related content of very high quality, and I will mention some of them in a later post.