- Joined
- May 8, 2002
- Posts
- 1,172,134
- Reaction score
- 59
You must be registered for see images attach
Chairs with March Madness logo are seen prior to the first round of the NCAA college basketball tournament, Wednesday, March 18, 2026, in Buffalo, N.Y. (AP Photo/Yuki Iwamura)
Copyright 2026 The Associated Press. All rights reserved.
Selection Sunday always brings debate. Each year, the NCAA selection committee blends résumé-based results with predictive analytics to build the bracket, creating inevitable gaps between how teams are seeded and how strong they actually are. This year, résumé-based metrics dominated seeding decisions. Predictive metrics have quietly proven their value in March. Last year, brackets built around KenPom projections would have landed in roughly the 97.5th percentile, underscoring how well efficiency-based predictive models capture true team strength compared to traditional résumé evaluation.
This analysis isolates that predictive signal. By averaging three widely used efficiency metrics, KenPom, Bart Torvik T-Rank and ESPN Basketball Power Index (BPI), a consensus ranking of team strength emerges. Comparing that ranking to the committee’s official order highlights where the bracket aligns with underlying performance and where it diverges, revealing the teams most likely to be overvalued or overlooked heading into the tournament. Those differences are where bracket value is often found.
March Madness Overseeding and Underseeding Methodology
To evaluate where March Madness seeding aligns with team strength, a dataset was constructed combining the committee’s full 68-team ranking with three widely used predictive metrics: KenPom, Bart Torvik T-Rank and ESPN Basketball Power Index (BPI). These systems are designed to measure how strong teams are on a possession-by-possession basis, independent of wins and losses. Each team’s ranking in those three systems was averaged to create a single composite measure of predictive strength. Because all three metrics are expressed as rankings, where lower values indicate stronger teams, the average preserves that ordering while smoothing out differences between individual systems.
You must be registered for see images attach
Duke forward Cameron Boozer (12) shoots against Virginia center Ugonna Onyenso during the second half of an NCAA college basketball game in the championship of the Atlantic Coast Conference tournament in Charlotte, N.C., Saturday, March 14, 2026. (AP Photo/Nell Redmond)
Copyright 2026 The Associated Press. All rights reserved.
Teams were then re-ranked from 1 to 68 based on this average, producing a “predictive ranking” that reflects how the field would look if seeding were based solely on underlying performance rather than résumé. This ranking serves as a baseline for comparison against the committee’s official ordering.
The difference between the predictive ranking and the committee’s ranking was calculated for each team. Positive differences indicate teams that were overseeded by the committee, while negative differences identify teams that were underseeded relative to their predictive strength. The results highlight a clear divide between résumé-driven selections and underlying team strength.
March Madness Overseeded And Underseeded Results
You must be registered for see images attach
Comparing predictive metrics to committee rankings reveals Iowa, Louisville and Tennessee as among the most underseeded teams, while TCU, Miami and North Carolina appear significantly overseeded.
Image created by author.
March Madness Overseeded Teams
Several teams were ranked significantly higher by the committee than their predictive profiles would suggest.
- TCU: TCU shows the largest gap on the overseeded side. The Horned Frogs were elevated by résumé-based metrics to No. 34. Predictive systems like KenPom and T-Rank rate them closer to the bubble at No. 42, suggesting their seed may overstate their underlying strength.
You must be registered for see images attach
LUBBOCK, TEXAS - MARCH 03: Xavier Edmonds #24 of the TCU Horned Frogs celebrates after a game against the Texas Tech Red Raiders at United Supermarkets Arena on March 03, 2026 in Lubbock, Texas. (Photo by John E. Moore III/Getty Images)
Getty Images
- Miami (FL): Miami’s placement reflects strong résumé indicators, but its efficiency metrics are more modest. The committee ranks them at No. 27 in the field, while the predictive metrics have them at No. 34.
- UCF: UCF benefits from a résumé that outperforms its predictive profile. While the Knights accumulated solid results, their efficiency metrics suggest a lower overall team rank (No. 45) than their seed indicates (No. 38).
You must be registered for see images attach
UCF forward Jordan Burks (99) dunks the ball during the second half of an NCAA college basketball game against Cincinnati in the second round of the Big 12 Conference tournament Wednesday, March 11, 2026, in Kansas City, Mo. (AP Photo/Charlie Riedel)
Copyright 2026 The Associated Press. All rights reserved
- UNC: North Carolina’s brand and résumé both play well in the selection process, but predictive metrics are less bullish. The gap between the No. 22 ranking from the committee and the No. 29 ranking by the metrics suggests the Tar Heels may be seeded closer to their résumé ceiling than their underlying performance.
March Madness Underseeded Teams
On the other side, several teams appear undervalued relative to their predictive strength.
- Iowa: Iowa stands out as one of the most undervalued teams. Despite strong efficiency numbers, the Hawkeyes’ résumé falls short of top-tier standards, leading to a lower seed. The committee has them at No. 36 compared to the No. 27 ranking using predictive metrics.
You must be registered for see images attach
Iowa guard Isaia Howard (23) drives past Michigan forward Morez Johnson Jr. (21) during the second half of an NCAA college basketball game, Thursday, March 5, 2026, in Iowa City, Iowa. (AP Photo/Charlie Neibergall)
Copyright 2026 The Associated Press. All rights reserved
- Louisville: Louisville’s predictive metrics suggest a significantly stronger team than its seed. Louisville is ranked by the committee at No. 23 but predictive metrics suggest a No. 14 ranking. This ranking does not account for the injury to Mikel Brown Jr.
- Tennessee: Tennessee combines strong efficiency metrics with solid but not elite résumé indicators. The gap suggests the Volunteers may be one of the more dangerous teams relative to their seed. They are ranked No. 21 by the committee and No. 13 by the predictive metrics.
You must be registered for see images attach
Tennessee guard Ja'kobi Gillespie (0) shoots a basket past Auburn guard Elyjah Freeman (6) during the second half of an NCAA college basketball game in the second round of the Southeastern Conference tournament Thursday, March 12, 2026, in Nashville, Tenn. (AP Photo/George Walker IV)
Copyright 2026 The Associated Press. All rights reserved
What This Means For Your March Madness Bracket
If predictive metrics are a reliable measure of team strength, then the gaps between those rankings and the committee’s seeding create opportunity. The bracket is built on résumé. Predictive metrics are built on future performance. Teams that rate higher in efficiency metrics than their seed suggests may be undervalued, while teams elevated by résumé may be overvalued relative to their true strength.
In practical terms, this creates a form of bracket arbitrage. The market, in this case the committee’s seeding, reflects one view of team quality. Predictive metrics offer another. When those views disagree, there is an opportunity to take advantage of the gap. That does not guarantee outcomes. March Madness remains highly volatile, and single-elimination formats introduce randomness that no model can fully capture. But if predictive metrics like KenPom and Bart Torvik T-Rank are directionally accurate, then the teams identified as underseeded represent some of the most logical places to look for upside.
Continue reading...
