2026 NCAA Tournament Bracket Projection
Generated Sunday, April 5, 2026
Field
Bids
Bids
Projected #1 Seeds
The current #1 seeds in our projected NCAA Tournament bracket are Duke, Michigan, Arizona, and Houston. According to our model, these teams have earned the top line with impressive resumes, led by Duke with a bracket score of 99.5. Duke's strong NET ranking of #1, combined with a 17-1 conference record and 17-2 mark in Quad 1 games, solidify their position as a top seed. Michigan, with a bracket score of 99.1, is close behind, boasting a 19-1 conference record and a 17-3 Quad 1 record.
Arizona and Houston round out the #1 seeds, with bracket scores of 98.0 and 93.3, respectively, according to our model. Arizona's 16-2 Quad 1 record and 16-2 conference record demonstrate their ability to perform against top competition. In contrast, Houston's profile is more nuanced, with a 10-6 Quad 1 record, but an unblemished 9-0 mark in Quad 2 games. While their NET ranking of #5 is the lowest among the top seeds, their overall body of work, including a 14-4 conference record, is sufficient to secure a #1 seed. Duke, Michigan, Arizona, and Houston have all demonstrated the consistency and strength required to be considered top seeds, and their respective resumes reflect their dominance in their conferences and against top-tier opponents.
The last four teams projected in the NCAA Tournament field are Tulsa, UCF, Texas, and Oklahoma. According to our model, Tulsa is holding on to a spot with a bracket score of 75.0, despite a lackluster quad 1 record of 0-1. The team's 13-5 conference record and NET ranking of 52 are sufficient to keep them in the field for now. UCF, on the other hand, has a slightly lower bracket score of 74.6, but its 5-8 quad 1 record and 9-9 conference record are enough to keep them ahead of the bubble. Texas and Oklahoma are in a similar position, with bracket scores of 74.5, and their quad 1 records of 6-9 and 4-10, respectively, are a concern.
Texas and Oklahoma are particularly vulnerable to being pushed out of the field due to their subpar conference records of 9-9 and 7-11, respectively. Oklahoma's NET ranking of 48 is also a concern, as it is the highest of the four teams. UCF's quad 2 record of 6-3 is a positive, but its NET ranking of 51 is only slightly better than Tulsa's 52. According to our model, any of these teams could be replaced if they suffer a bad loss or if another team makes a strong case for inclusion. Tulsa's lack of quad 1 wins is a major concern, and if they do not improve their resume, they could be the first to fall out of the field. Texas and Oklahoma need to hope that their quad 1 wins are enough to overcome their conference records and keep them in the tournament.
The first four teams out of the NCAA Tournament field are New Mexico, SMU, San Diego State, and Auburn. According to our model, these teams have bracket scores of 74.3, 73.9, 73.8, and 73.5, respectively. New Mexico needs to improve its Quad 1 record, currently sitting at 2-7, to bolster its tournament resume. With a NET ranking of 46, the team must focus on strengthening its overall profile. San Diego State, on the other hand, boasts a strong conference record of 14-6, but its Quad 1 record of 3-8 is a concern.
SMU and Auburn face similar challenges, with both teams struggling in Quad 1 games. SMU has a Quad 1 record of 4-9, while Auburn's is 4-13. To play their way into the tournament, these teams must close the gap in their resume by performing better in top-tier games. According to our model, Auburn's bracket score of 73.5 indicates that the team needs to make significant improvements to its profile, which is currently hindered by a 7-11 conference record. San Diego State, with a NET ranking of 47, must also work on enhancing its Quad 1 performance to move up the bracket rankings. New Mexico, with its 13-7 conference record, needs to build on this success and demonstrate its ability to compete against stronger opponents.
The current state of the bracket remains largely unchanged at the top, with Duke, Michigan, Arizona, and Houston still holding onto their number one seeds. According to our model, these teams have maintained their strong positions, with no significant movement in their bracket scores. Notable trends include the addition of Oklahoma to the bubble as one of the last four teams in, replacing New Mexico which has dropped out of consideration. The overall field size of 68 teams, comprising 31 auto-bids and 37 at-large selections, remains unchanged. As the season continues to unfold, the stability at the top of the bracket is a notable aspect, with the top seeds solidifying their positions and the bubble teams jockeying for position in the remaining at-large spots.
How Our Bracket Model Works
Normalized 0–100 from rank position. The NCAA's own evaluation tool combining wins/losses and game-level efficiency across all Division I opponents.
Weighted quality score — Q1 wins +5, Q1 losses −1, Q2 wins +2.5, Q2 losses −2.5, Q3 wins +0.5, Q3 losses −5, Q4 wins 0, Q4 losses −8. Normalized 0–100.
SoR rank normalized 0–100. Measures how impressive a team's record is given the difficulty of its schedule — a 20-win team in a weak conference scores lower than a 20-win team in the ACC.
Adjusted offensive minus defensive efficiency (points per 100 possessions). Captures how dominant a team is regardless of pace. Normalized 0–100 across the field.
60% road record value + 40% SOS rank, both normalized. Rewards teams that schedule tough and win away from home — factors the committee explicitly values.
Final bracket score = weighted sum of all five components, scaled 0–100.
Our Model vs. The Selection Committee
The NCAA Selection Committee uses the same core inputs — NET rankings, quad records, strength of schedule, and road record — but applies subjective judgment to each case. Committee members can weigh injuries, recent form, head-to-head results, conference tournament performance, and what is often called the “eye test.”
Our model is purely data-driven: the same formula applied consistently to every team, with no adjustments for narrative or circumstance. That removes human bias — but it also means we can't account for context that only humans can evaluate. When the model and the committee diverge, it's often because of factors that don't yet show up in the numbers.