Grading the Committee – 2025 Edition

For last season’s write-up, see here.

The 2025 bracket is out. The Kansas Jayhawks earned a 7-seed, fitting in where most bracketologists had them. Most KU fans are fine with this seed as they believe the Hawks didn’t have a good enough season to deserve to be any higher.

Fans of other teams are not so happy, and there is some controversy about the inclusion of North Carolina in the field (and exclusion of West Virginia).

The committee has the difficult task of selecting 37 at-large teams and seeding 68 teams in a way that is fair and impartial. They have to sort through dozens of resumes in a relatively short period of time and, in the words of one bracketologist, often compare apples to oranges. Teams each have different schedules, records, and strengths that make them difficult to compare. How much importance should be placed on metrics? How much emphasis should be placed on results against the very best (Q1A games) or avoiding big upsets? How should SOS be incorporated?

These questions are all worthy of debate, but what we feel is most important is that, whatever the committee decides is the proper way to weight the data on the team sheets, it is consistent across the field of teams it analyzes. In other words, if the committee is very concerned with a team’s results in Quad 1 games it can’t overlook when a team performed poorly in this area. Or, if metrics are important, they should be applied the same across the board. Rewarding one team for having great computer metrics but not another isn’t fair.

For the 2025 season, we broke down the team sheet into four broad categories which were then quantified so teams could be ranked in each category. We then applied weights to these categories so as to best capture how the committee emphasized the categories. These categories, and corresponding weights, are:

• Schedule Strength – 2.0%
• Overall Winning Percentage – 2.2%
• W-L Results in Quadrants – 40.1%
• Computer Metrics – 55.7%

More detail on how these weights were arrived at (and other assumptions) will be provided at the end.

In applying the weights this way, we get the closest correlation between how a mathematical model would rank the teams and the final S-curve itself. What we are doing is testing the committee’s consistency.

Last season, the closest correlation we got was an R² of 0.8793. For 2025, this jumped to 0.9498. Now, this could partially be due to us doing better at applying the formula for finding the proper category weights (the process is a trial-and-error one to find the closest tie between the committee’s S-curve and our own), but suffice to say the committee did better overall in 2025 than it did in 2024.

Still, the 2025 bracket wasn’t without flaws or controversy. We’ll go through the teams most affected.

Memphis

The Tigers were AAC Champs both in the regular season and conference tournament, finishing the season at 29-5 overall and earning a 5-seed. This wouldn’t seem all that surprising, except that the Tigers played the 90th best overall SOS, had 2 Quad 3 losses, and an average computer metric of 37.1. These poor marks tended to bring other teams down (see Drake or UC San Diego), but not Memphis. While the committee had Memphis as the 20th best team on the S-curve, we had them at #31, a difference of 11 spots and the difference between a 5-seed and an 8-seed.

West Virginia

The Mountaineers were the committee’s first team out of the field, ranking #47 on the S-curve. We had them at #36, which would have not only included them in the Big Dance it would have given them the final 9-seed and would have bypassed them from Dayton. They were included in all 111 final brackets on Bracket Matrix. Their exclusion was the biggest shock of the night.

West Virginia didn’t have a single bad loss and finished 10-13 in Q1+Q2 games…which one would think was good enough to make the field. The committee did claim that they dinged WV for not having their best player, but Tucker DeVries was absent for all but 8 games of the season and the Mountaineers had big wins against Iowa State and Kansas without him.

Three Other Under-seeded Teams

Michigan, the Big Ten Tournament champions, was 17 on the S-Curve (5-seed) but had an argument to be #11 and a 3-seed. The Wolverines being placed so low was yet more proof that the committee fails to properly account for conference championship games (particularly those played on Selection Sunday).

Louisville, the ACC Tournament runners-up, is an 8-seed and at #29 on the S-curve. We have them as #23 on the ranking system (6-seed). The Cardinals do get to play in Lexington, however, but their spot in the field is suspiciously poor given their resume. Flipping Louisville with Memphis on the S-curve would substantially improve the consistency of the committee’s ranking.

Utah State is in the field as a 10-seed (#40 on S-curve) but should be a bit higher (#34 in our ranking). No bad losses and won the only Q1A game it played. From a position-in-the-bracket aspect, this may actually be better for the Aggies as the 10-line misses out on playing a 1-seed until the Elite 8 at earliest.

Let’s Talk North Carolina

The Tar Heels made it into March Madness as the last team in the field and will play in Dayton against San Diego State. Their inclusion was derided for a few reasons. One, their 1-12 record in Q1 games was seen as evidence that they are not good enough to beat tournament-quality teams. Two, the committee chair this year was UNC’s athletic director. While by rule he does have to leave the room when his team is being discussed, it’s tough to deny that he still has influence on the process overall. Three, people see this as evidence that name brand matters in selection, and there isn’t any names bigger in college basketball than North Carolina.

As far as committee consistency goes, UNC was our 50th ranked team (46th on the official S-curve). This put them behind not only West Virginia, but also non-Tourney Indiana and Ohio State. It’s really tough to defend UNC’s place in the field if we are being consistent.¹

While UNC’s 1-12 record against Q1 opponents is often mentioned, what gets overlooked is its Q3 loss to Stanford in January. Bad losses have held teams out in the past, and this season we found that the committee applied weights to them as well…just not for North Carolina².

Xavier – Surprisingly Strong

Xavier was another bubble team with poor Q1 results that made it into the field…going 1-9 in Q1 games. The Musketeers got in at #42 (still an 11-seed play-in team). We had them at #46, which is a bit below the committee but still in the field. Xavier was a surprise, but they weren’t necessarily a poor inclusion.

Should be In, Should be Out

West Virginia for North Carolina for reasons already stated. Other than that, the committee didn’t really snub or rescue anyone else (though it did over and under-seed certain teams as we’ve discussed).

Bracket Matrix

There were 225 Final brackets listed on Bracket Matrix, and I’m assuming that these were all published before the field was announced. Using the average ranking and number of brackets a team made it into, simple regression analysis was done to compare the Bracket Matrix ranking (effectively what the consensus S-curve was) to the NCAA’s actual S-curve. This was a correlation of 0.9595. This is slightly more-closely correlated than what we could get our ranking to show.

For that reason, we have no reason to suggest that Bracket Matrix isn’t the best predictor of what the field will end up looking like. Still, for future seasons we will use the team sheet data as an independent predictor of a team’s seed.

The next correlation we ran was between the Bracket Matrix consensus ranking and our own…this was at 0.9741. So, even not fully knowing what the committee would value as the most important criteria, the consensus ranking was more consistent with the ranking than the committee itself.

Some have suggested that the committee, currently made up of athletic directors and conference commissioners, should be replaced with full time bracketologists, with many even wondering how the committee is allowed to have such conflicts of interest.

This struck me as ironic. Historically, the NCAA Tournament selection committee has always been made up of AD’s and conference commissioners, as the institutions and conferences just are what make up the NCAA. As someone has to make decisions as how to best field a tournament, naturally it was determined that a number of members from a variety of schools (both big and small) and conferences (geographically distinct) would be the fairest way.

Back then there were no “bracketologists” nor any advanced metrics that could determine with more precision and accuracy what the “fairest” ranking would look like. The committee tried its best and the tournament was played with little controversy.

It is the popularity of the Tournament and its importance which made making the Field so important, and eventually Joe Lunardi made the discovery that college basketball fans were extremely interested in knowing what the Field might look like before Selection Sunday, becoming the first ever “bracketologist” and starting the trend which has grown to where it is today³.

That’s what’s ironic. A bracketologist is not intended to be someone who selects or ranks the field, he is rather someone who predicts how the committee will select and rank the field. If we remove the committee and replace it with a group of bracketologists, what we would be doing is actually picking a field on how a group of people think a hypothetical committee would pick a field…essentially replacing a substance for an ethereal idea, an object for its shadow. Yet, at least this season, such an exercise would produce a more-consistent bracket.

The S-Curve and Matchups and Conspiracy Theories

The biggest thing the committee does is set the S-curve. In order to get to 68 teams, the committee takes 31 automatic qualifiers and the best 37 at-larges and then ranks them, 1 to 68, to produce a field. The next step, placing the teams in appropriate seeds and match-ups is a relatively quick process. But it is the first aspect that the committee is most concerned with. It doesn’t want to include a team that doesn’t belong or grossly over or under rate someone.

Fans, on the other hand, tend to look at the process the other way. A fairly large portion thinks the committee sets matchups to create storylines, often attributing financial motivation to their decisions. But this just isn’t how the committee operates. The matchups are often an afterthought and are set up normally based on geographic reasons or other bracketing principles, such as conference affiliation. For instance, UConn, an 8-seed, couldn’t have played Creighton, a 9-seed, as both are in the Big East and have played each other multiple times this season. So a bracket that had them in different regionals was the one they made. (Similarly 8-seed Mississippi State had to play 9-seed Creighton or Baylor as the other 9’s are SEC teams).

But one area where the committee does need to focus on matchups, certainly in the eyes of many coaches, is proximity to a pod or regional site for worse-seeded teams. 8-seed Louisville gets to play down the road in Lexington, where should it make it to the Round of 32, would likely face 1-seed Auburn. Wouldn’t it benefit Auburn to play another 8/9 pairing, or at least move the regional somewhere else (such as Wichita)? Maryland fans are upset their 4-seed Terps have to go out to Seattle (where they face a Western team in 13-seed Grand Canyon). Wisconsin is a 3-seed and goes to Denver to face 14-seed Montana and potential 6-seed BYU. Such disadvantages persist for better-seeded teams across tournaments. This is partly due to the NCAA wanting to reduce travel costs and burdens (a noble goal), but if it harms the better seed, how fair is it?

Conclusion

The committee likely did better selecting and seeding the field in 2025 than it did in 2024, but it still wasn’t without flaws. The biggest error was Memphis being a 5-seed when the weighted ranking method concluded they should have been an 8. There was only one real snub/false inclusion this year, with West Virginia being left out for North Carolina.

The consensus bracket on Bracket Matrix was more consistent than the committee itself even using the committee’s implied preferences, and even with a committee that can seem schizophrenic, the final consensus on BM was very predictive of what the field actually was.

1. Ironically, North Carolina’s ranking in the purest-resume metric, WAB, had them in the field and at a safer #42. So there is justification for North Carolina being included. The problem is that one can make this argument with most bubble teams, simply find the best part of the team’s resume and highlight that. The point is that if the committee shows it considers all aspects of the team sheet (it does) and if we think the committee should be consistent in how it applies this data (we should), then UNC should be out of the field. ↩︎
2. If we removed UNC’s bad loss to Stanford, this would have bumped the Heels from #50 in our ranking to #46…matching the committee. So, by effectively ignored this Q3 loss, the committee put UNC in the tournament. ↩︎
3. https://en.wikipedia.org/wiki/Bracketology ↩︎