The 2018 NCAA men’s basketball tournament field will be announced Sunday. According to a new study by researchers at the University of Illinois, you might not have to wait that long to know who’s in or out.
In their paper, “Modeling the NCAA basketball tournament selection process using a decision tree,” computer science professor Sheldon H. Jacobson and former graduate student Shouvik Dutta created a decision-tree model to simulate the process used by the tournament’s selection committee.
“The committee has a well-defined set of criteria that they use in making their picks,” Jacobson said. “There is always a certain amount of confusion over which teams — especially those on the bubble — will be selected and which will be left out.”
Jacobson and Dutta studied teams who were 10 seeds or higher in past NCAA Tournaments as well as No. 1 and No. 2 seeds in the National Invitation Tournament, which are often the teams who just missed the NCAA Tournament.
Sign Up and Save
Get six months of free digital access to the Ledger-Enquirer
Using the same data considered by the selection committee, Jacobson and Dutta created a step-by-step process by which they dissected teams in pairs to determine which teams were the most dominant. The teams with the most pair-wise victories were included in the tournament field.
Their 12-step process focused on each team’s RPI rankings, Pomeroy ratings, strength of schedule and the teams’ win-loss records in their last 12 games. Between 2012 and 2016, their model predicted 90 percent of the bubble teams — those on the verge of making or missing the tournament — correctly.
As far as this season’s bubble teams, their model predicts St. Bonaventure, Utah, Oklahoma State and Saint Mary’s are the Last Four In. Teams they have on the outside looking in are Alabama, Marquette, Baylor, Providence, Mississippi State, Syracuse and LSU.
Jacobson pointed out that while the model is highly accurate, there are typically exceptions.
“Every year, there has been one team that we select that the committee does not,” Jacobson said. “This suggests that, even with all the data available, there is a certain amount of human input and uncertainty that goes into the selection process.”