NCAA Tournament’s early bracket reveal shows selection process remains as much art as science
If you’re expecting the NET to catch up with your favorite collegiate basketball team, don’t be surprised if a tear sheds from your eyes.
When the NCAA Tournament Selection Committee did its annual Bracket Reveal preview, about three weeks before Selection Sunday (and a far cry from the attention swallowed up by the Super Bowl), all 16 participating teams were in the NCAA’s top 25 Scoring tool metric that is at the heart of the selection process.
But the four saplings per region allocated this week – they are the saplings as they would be today and will likely change in the remaining games – were not identical to the NET leaderboards or other metrics reported on the Use the members’ “Team Sheets” to guide their decisions.
PREDICTING BRACKET: Bill Bender projects the Field of 68
For example, Alabama, ranked #1 overall, is second in the NET and in prediction ratings compiled by statisticians Ken Pomeroy and Jeff Sagarin, and third in ESPN’s BPI. The Crimson Tide is #1 in both outcome-based metrics, analyst Kevin Pauga’s KPI and ESPN’s Record Strength. But if you think these are more accurate ratings, you’ll notice that SOR #14 Tennessee is ninth overall at the unveiling of the bracket, and SOR #17 Iowa State is ranked six spots higher by the committee at #11.
Most importantly, what the committee has told us is that while there are mathematical elements in the NCAA tournament seeding process, it’s still more art than science.
MORE: Breakdown of updated odds for 2023 NCAA tournament
The Committee has always insisted that the NET – and its predecessor, the Ratings Percentage Index (RPI) – are tools used by the Committee to group teams for the purpose of rating their resumes. In the old RPI-based system, the selectors recognized teams that achieved wins against opponents who were in the top 50 or top 100 of the RPI leaderboard.
However, at a January 2017 meeting, Pomeroy, Sagarin and others persuaded the NCAA that they did not accurately grasp the increasing difficulty of winning street games or games on neutral courts as opposed to home games. The math showed that beating the 75th team on the road was roughly equivalent to beating the 30th team at home. This led to the formulation of the quadrant system.
MORE: Powerful Big 12 has a chance to set March Madness records
The NCAA, aware that the RPI metric was also flawed, decided to develop a new metric. Although the NCAA insisted that a team’s margin should play only a minor role, the NET has been found to align very closely with outcome-based metrics, particularly KenPom. Many fans are convinced that a high NET ranking automatically leads to a tournament spot or favorable seeding for the team they follow.
But Rutgers made it into the field last year with a NET ranking of 77. In 2019, NC State was disqualified with a NET of 33.
It’s not a magic formula. It’s a tool. Committee members have insisted since the NET was introduced for the 2018-19 season, as they have for years regarding the RPI. Just as it’s not easy to win enough games to make the NCAA field, neither is it easy to look at computer ratings to determine if a team is worthy of being included in the field.
Here were the seeds the committee unveiled during a half-hour broadcast Saturday on CBS, with their position on the NCAA’s 1-16 seed list in parentheses:
Region East
1. Puree (3)
2.UCLA (8)
3. State of Iowa (11)
4. Markette (14)
Region South
1. Alabama (1)
2. Baylor (7)
3.Virginia (10)
4. Indiana (13)
Middle West
1.Houston (2)
2.Texas (5)
3. Tennessee (9)
4. Xavier (16)
West region
1. Kansas (4)
2. Arizona (6)
3. State of Kansas (12)
4. Gonzaga (15)
Some little surprises and some things we learned:
– Arizona, which earns No. 2 in the West over UCLA, shows the committee values performance against top opposition more than statistical excellence. Arizona is No. 11 in NET and KenPom (due at least in part to a less-than-stellar defense affecting its predictive metrics) and No. 8 in strength record (due to losses to four opponents likely to miss the tournament) . . But the Wildcats own wins over three of the top 16 seeds, including the Bruins. UCLA was better analytically (#3 in KenPom, #5 on the NET) but has not beaten anyone ranked higher than #7 in the composition of 94 online brackets at BracketMatrix.com.
— The committee estimates the performance of the Big 12. Among the seven best teams, three were members of the conference, and five presented the top 4 seeds. That includes Kansas State, which earned a #3 spot despite losing four of its last five games, including to Texas Tech and Oklahoma, neither of which are included in a projected box in the Bracket Matrix.
– The closeness of the seed rankings for Kansas (fourth overall) and Texas (fifth) is an indication that they are almost neck and neck for a top-row spot, although Baylor (seventh) may also have a chance to make his mark assert yourself in the competition. The Bears took on each of them with a game on Saturday. KU and Texas play again on March 4th, the final day of the regular season for both.
– For those still believing the novelty is having an impact, Xavier earned a No. 4 seed in the Midwest region despite back-to-back losses at sub-.500 Butler and Big East leader Marquette. That’s why the three-time loss in four games didn’t knock Purdue off the No. 1 seedline; The Boilermakers were still among the four teams most valued by the committee.