Saint Mary's and the danger of seeding teams by strength of record
Which teams can disrupt the bracket?
The 2017 NCAA tournament brackets are set, and, given the layout of the field, the men's basketball selection committee once again emphasized the value it places on strength of record in the seeding process. Though I'm not suggesting the committee actually uses SOR, comparing this year's seeds with the current SOR rankings does show that, when assigning seeds, this committee places high importance on evaluating how impressive a team's won-loss record actually is.
As I've argued previously, however, the committee might benefit from a shift in thinking there. While it's fine to use strength of record to select teams for the tournament, if the committee wants the best event, seeding should be about how good the teams actually are.
That's where a metric such as the Basketball Power Index (BPI) should come into play.
Saint Mary's provides a great case study for the flaws in the committee's current seeding strategy. While the Gaels have won a lot of games, they have not had many "good wins" outside of beating Dayton on the road -- and their strength of record of 25 reflects as much. Analysis of the committee's decision-making suggests that the team's 7-seed has everything to do with that aspect of its résumé.
But Saint Mary's has played better than a 7-seed. A BPI-powered analysis of every game shows that the Gaels' efficiency -- adjusted for a variety of factors such as opponent and game location -- makes them the 12th-best team in the country, a 3-seed if seeding were based purely on BPI. BPI's predictive component says Saint Mary's is in good position to outperform a typical 7-seed, and the seeding in the West amplifies the possibility and outlook for the Gaels in this tournament.
No. 7 seeds are usually favored to win their first game, and BPI gives Saint Mary's an 82 percent chance of beating VCU in the first round. Assuming Arizona wins its first-round game, as well, that sets the Gaels up for a major upset -- according to the seeds, anyway. But BPI says Saint Mary's (12th) is better than Arizona (24th), and gives the Gaels roughly a 55 percent chance to win that game and move on. In fact, after simulating the tournament 10,000 times, Saint Mary's is actually the ninth-most likely team to win the tournament at 3.5 percent -- well ahead of 3-seeds such as UCLA (2.2 percent), Baylor (1 percent) and Florida State (1.1 percent).
The team that suffers in this (besides Saint Mary's, of course) is Arizona, which is not going to face a typical 7-seed in its second-round game. The other three 7-seeds in this field have an average BPI rank of 32, meaning the Wildcats' road to the Sweet 16 is more challenging than that of any other 2-seed.
Does it make for a better tournament when either a 2-seed or a team that is top-12 nationally, based on predictive metrics, is certain to get bounced in the round of 32? That's a question the committee should confront sooner rather than later.
For more from ESPN Analytics, visit the ESPN Analytics Index.