Modeled Predictions for Selection Sunday

Mar 12, 2015; Chicago, IL, USA; Indiana Hoosiers head coach Tom Crean reacts during the first half in the second round of the Big Ten Conference Tournament against the Northwestern Wildcats at United Center. Mandatory Credit: Jerry Lai-USA TODAY Sports

At 6PM ET today, the NCAA selection committee will announce the 68 particulate in the 2015 tournament. 27 teams are already guaranteed spots by virtue of automatic bids, where the last 5  automatic bids will be decided later today in conference finals (A10, AAC, Big10, SEC, Sun Belt). Over the course of the last several weeks, I have been tracking the status of each team’s probability of making the NCAA tournament, based on their remaining schedules, projected conference tournament results, and their chances of an at-large bid as decided by an at-large bid predicting model that I developed based on the last five years of committee tournament selections. In addition to measures that account for the actual ability of teams (efficiencies, strength of schedule, etc.), the bid predicting model also takes into account slightly more dubious factors the tournament committee considers when making selections, factors like RPI and AP preseason ranking.

The original post detailing the methodology behind this method can be found here.

To account for the 5 games still taking place today, I ran those matchups through a basic college game model that I have developed. This simple model predicts games on factors like efficiency and strength of schedule and produces percentage projections of each team’s chances of winning, as well as projected spreads, detailed below.

For the 10 teams above, their percentage projections are factored into their overall tournament probability. All other conference bids have already been decided. The teams listed below are 100% in by nature of their automatic conference bid.

All teams not yet mentioned will have to rely on the strength of their at-large bid chances alone. What differentiates this procedure from most other bracketology exercises is the ability to put an exact number on each team’s chances. Below are those numbers, sorted by overall chances of a tournament appearance, and then by a team’s chance of an at-large bid.

Based exclusively on these numbers, I predict the last four teams in are Georgia, Mississippi, Florida, and Indiana. The first four out are Nebraska, Oklahoma St., Miami (FL), and Purdue. Expect an interesting selection Sunday though. This model doesn’t think any team past 63 has more than a 50% chance of getting in, which means it predicts more teams than usual have a decent chance of dancing.