Each year, 68 teams are invited to participate in the NCAA basketball tournament. Of these 68 teams, 36 are given an at-large bid by a selection committee rather than automatically qualifying. However, despite the process being public, the exact method by which teams are ranked against other teams to determine which teams are given at-large bids is not known. Many factors are known or theorized to be in use, including the RPI, top 50 wins, and strength of schedule. In this paper we attempt to provide a quantitative method to mimic the team made by the selection committee by constructing a decision tree. We then test our method on years 2012-2016, where we find that in each year all but one team selected by our method was invited to the tournament.
© Copyright 2018 Journal of Sports Analytics. IOS Press. All rights reserved.