Bracket Pulse.
Sixty-eight teams, 31 automatic bids and 37 at-large, seeded on a full S-curve and placed into four regions. The 2026 tournament is finished, so this page shows the model’s own board as it stood on Selection Sunday against the bracket that actually played out. Michigan won it, having been given 20% on the morning of the draw. The 2026-27 projection appears here once that season is simulated.
Every percentage on this page is the model’s chance of a team reaching that round at all, published on March 15, 2026 before a game was played, not conditional on any bracket. Every winner and score is the recorded result. Those two are deliberately shown together, because the point of a settled bracket is the comparison. The file this page used to read for its odds column now carries the finished tournament rather than a forecast: across all 68 teams every published value is exactly 0 or exactly 1, which is an outcome, not a probability. Those flags are used here only for what they really are. A conference’s champion takes its automatic bid before any at-large place is filled, which is why a 15 seed sits in the field while a better team misses. Four teams on lines 11 and 16 played into the bracket in Dayton, listed in the sidebar.
The Final Four
Indianapolis · percentages are the model’s odds at the drawUConn won the region at 7%, outside the model's top three here.
Arizona won the region at 42%.
Illinois won the region at 18%.
Michigan won the region at 45%.
The projected S-curve
68 places · four regions assigned after the lines are setLines 1 through 4 are the protected seeds, which is the line worth arguing about; from line 11 down, most places belong to a conference’s automatic bid rather than to an at-large resume. Two cells hold a pair: the four FF games on lines 11 and 16 played into the bracket in Dayton.
Title contenders
all 68 teams · odds as of March 15, 2026Each column sits strictly below the one to its left, because these are chances of reaching a round at all and the fall between them is that weekend’s attrition. Title odds are flatter here than in any other sport on the site: six rounds of single elimination against a 68-team field held even the model’s strongest team under 20%, where a No. 1 seed in football sits near 28%. That is the tournament, not the model. Result is the round the team actually reached.
The mock carries three more columns this sport cannot fill: proj W-L needs a season simulation, win conf needs a conference for each team and no NCAA feed here records one, and make 68 would read 100% for all 68 rows in a field that is already settled. They are left out rather than filled with a number that means something else, which is also why there is no auto-bid watch card.