Our sports and fandom analytics projects
The core of the Fanalytics Project is a program of applied research sponsored by the Emory Marketing Analytics Center. This research program focuses on a wide variety of sports and fandom analytics projects. The projects range from league level analyses of how revenue sharing influences competitive balance to studies of the importance of mascots.
The projects are always applied in nature and executed using data and advanced analytics techniques. Techniques range from simple linear regression to dynamic optimization and machine learning. We also make it a point to ground our work in economic and psychological theory. This is an important distinction because for analytics to have an impact the models need to be consistent with how humans make decisions.
The research is reported in a variety of formats. In addition to a brief overview of the project that includes contextual details and key insights we also provide academic publications, working papers or popular press versions.
Examples are abundant. Teams “tank” to get better draft choices. Tanking occurs because losing more games results in getting better future talent. Teams cut veterans and go with younger players who are not yet eligible for free agency. Teams shift to rookies because these players have less bargaining power and are cheaper.
The research linked to on this page reports the results of analyses related to revenue sharing in Major League Baseball. The research was conducted at a time when the league was concerned that the Yankees would dominate the league by out spending other teams. While this fear has faded it is a classic issue in sports – teams with better brands and revenue potentials can outspend small market rivals.
Baseball’s answer to this was to develop a system of revenue sharing that penalized teams for exceeding some “luxury” threshold. The problem was that this system created an incentive for teams to become less competitive on the field. Again, it is a simple idea. If you create an incentive for poor performance (transfer payments based on local revenues or better draft picks for losing more games) you will get more “poor” performance. The problem with such a system is that you can reduce competitive balance. Competitive balance matters if you believe that it’s important for every team to have a chance at being a winner.
To study the issue, I created a structural dynamic programming model of team’s decision making. This is a statistical model that captures how dynamic concerns such as building a fan base and short-term preferences for winning influence team’s decisions of how much to invest in payroll.
The results suggest that the key to avoid “tanking” is to base revenue sharing on winning and population rather than on local revenues. In other words, create incentives for being competitive rather than incentives for losing.
The research is reported in two forms. The first is a link to a New York Times piece that explains the results to a general audience. The second is a link to the academic paper in the Journal of Marketing Research. This one provides the full details and shows some hardcore sports analytics.