Features / NCAAB

---

Two weeks before Selection Sunday and the BXS model is telling a story the bracket doesn't show yet. Three teams share 28-2 records but their trajectories have diverged sharply. A popular 2-seed has been sliding for two straight weeks. And the biggest BXS movers in the country aren't the names your bracket pool is talking about.

This is the March edition of NCAAB Wrapped. Conference tournaments start next week. Selection Sunday is nine days away. The question that matters right now isn't "who's the best team?" It's "who's getting better?"

Three 28-2 Teams, One Clear Surge

Duke, Michigan, and Arizona all sit at 28-2. All three have BXS ratings above 1840. On paper, they look interchangeable. The two-week BXS movement says otherwise.

Source: Boxscorus BXS model, data through March 7, 2026.

Duke's +48 is the largest two-week BXS gain among the top 10 teams in the country. To put that in context, +48 is roughly the BXS gap between the 5th-best team and the 20th-best team in the country. Duke compressed that entire gap in two weeks. That's not just winning. That's winning by margins that impress the model, which weights blowouts more heavily through a margin-of-victory multiplier. A +48 move means Duke has been beating teams by more than the model predicted, game after game. Michigan gained +32, a solid stretch, but Arizona's +36 is steadier: six consecutive wins with rising BXS after each one, no dips.

The gap between Duke and Michigan is only 6 BXS points, which translates to roughly a 51-49 edge on a neutral court. But Duke is the team that's been accelerating. If these two meet in the Final Four, that 6-point gap could be 15 by then.

Houston and Purdue: Falling at the Worst Time

Not every contender is trending the right direction.

Source: Boxscorus BXS model, data through March 7, 2026.

Houston is a 2-seed that the model sees as a low 3, not the 2 the committee will give them. A -20 BXS swing in two weeks means they've been losing games outright or winning by margins far smaller than expected. For context, 20 BXS points is roughly a 3 percentage point shift in win probability against an average opponent. That doesn't sound like much, but in a 68-team tournament where the margins between a Sweet 16 run and a first-weekend exit are razor-thin, it changes everything.

The model doesn't know about injuries or fatigue or locker room dynamics, but it knows the results. Houston has been underperforming its own rating for two straight weeks.

Purdue's -20 tells a similar story. They've been bouncing between solid wins and deflating losses, the signature of an inconsistent team. In a single-elimination tournament, inconsistency kills you. One bad night and you're watching from home.

Compare these to Saint Mary's (1764 BXS), which gained +45 over the same period. The Gaels climbed steadily through late February, the kind of trend you want to see from a team entering March.

The Biggest Risers: Where the Real Stories Are

The top of the BXS standings gets all the attention. But the most interesting two-week movement happens further down the rankings, where teams are either locking up at-large bids or getting hot at exactly the right time.

Source: Boxscorus BXS model, data through March 7, 2026.

Dayton's +62 is the largest move on this list. A 21-9 record doesn't scream "tournament team," but context matters. The Flyers have been winning convincingly enough against A-10 opponents to jump 62 BXS points in two weeks. At rank #50, they're firmly in the at-large conversation. The committee will look at the resume. The model says the recent performance is real.

Cincinnati at +49 is a different kind of story. At 17-13, they look like a sub-.500 also-ran, but they play in the Big 12, where losses come against teams rated 1750+. A 49-point BXS surge means they've been winning the games they should and competing in the ones they shouldn't. The model's opponent quality factor (which scales K-factor based on opponent BXS) rewards this: beating a 1700-rated Big 12 team earns roughly 1.13x the normal K, while beating a 1300-rated team earns only 0.87x.

The Biggest Fallers: Who Should Be Nervous

Source: Boxscorus BXS model, data through March 7, 2026.

Liberty is the one that jumps off the page. If you just look at their record, 24-6, you'd call them a tournament lock. Your bracket pool will have them winning a game or two. But the model sees something different: a team that has dropped 52 BXS points in two weeks, meaning they're losing games outright or winning ugly against opponents the model expected them to handle comfortably. A 24-6 record in C-USA with a collapsing BXS trend is exactly the kind of profile that gets a team a 14-seed and a first-round exit. The record says "safe pick." The trend says "stay away."

BYU at -42 is a bubble team sliding in the wrong direction. At 20-10, they're on the edge. The committee uses the NET rankings and quad-system, not BXS, but the signal is the same: this team peaked three weeks ago and has been declining since. The model would tell you to bet the under on BYU's tournament seed if you could.

The Conference Power Ranking

The model's conference-level view reinforces what everyone already suspects about college basketball in 2026:

Source: Boxscorus BXS model, data through March 7, 2026. Averages computed across all D1 members of each conference.

The Big 12, SEC, and Big Ten are separated by just 7 BXS points at the conference level. That's essentially noise. The real gap is between the top three and everyone else: a 52-point drop to the ACC and 57 to the Big East. This gap shows up in tournament results. Teams from the top three conferences have had their BXS ratings stress-tested against each other all season. Teams from the ACC and Big East have had an easier path to inflate their numbers.

One thing the averages hide: the Big Ten has 18 teams, which means more low-end programs pulling the average down. Their top-end talent (Michigan at 1865, Michigan State at 1760, Illinois at 1761, Purdue at 1748) is as strong as any conference. But they also carry teams like USC, which dropped 67 BXS points in two weeks and sits at rank #160.

The Miami (OH) Question

Miami (OH) is 30-0. The only undefeated Division I team in the country. Their BXS is 1745, which ranks them 14th nationally, and they gained +19 over the last two weeks.

So why isn't an undefeated team ranked higher?

Because they play in the MAC. The model's conference prior for the MAC starts around 1460, meaning Miami had to climb from a lower baseline. And while going 30-0 is extraordinary, the opponent quality factor means each MAC win earns fewer BXS points than a win in the Big 12 or SEC. Miami's average opponent BXS is well below the national average, so their wins, while consistent, are individually worth less.

Is 1745 too low for an undefeated team? Maybe. The model is honest about what it knows and doesn't know. It knows Miami beats everyone they play. It also knows the teams they play are, on average, not very good. In the tournament, they'll face opponents rated 100+ BXS points above anyone on their regular-season schedule. The model gives them a 4-seed, which feels about right: good enough to beat a 13, favored against a 5, but an underdog from the Sweet 16 on.

I'll be honest: I'm not sure the model handles a team like this correctly. An undefeated record signals something about consistency and mental toughness that BXS can't directly measure. If Miami (OH) makes a run, it'll be the most interesting test case for the model all tournament.

What This Tells You About Your Bracket

Conference tournaments will reshuffle some of these trends. A team on a -40 slide can win three games in three days and enter the NCAA tournament with renewed confidence (and a 20-point BXS boost). But the trends from the last two weeks are the best leading indicators we have.

A few principles from the data:

Trust the surges. Duke (+48), Saint Mary's (+45), and Akron (+44) are peaking at the right time. The model doesn't know about "clutch" or "momentum," but it knows these teams are beating opponents by wider margins than expected. We suspect (but can't yet prove with our data) that late-season BXS surges correlate with deeper tournament runs. It makes intuitive sense: teams that are improving in late February have figured something out. But we'll need a few more years of tournament data to say that definitively.

Fade the fallers. Houston (-20) and Purdue (-20) are both popular bracket picks because of their names and seeds. The model says they've been declining for two weeks. That doesn't mean they'll lose in the first round, but it means their current BXS is lower than their reputation suggests. If you're looking for a contrarian pick, taking the field over Houston in the East region isn't crazy.

Watch the mid-majors. Miami (OH), McNeese (+37), and High Point are all trending up. Mid-major teams that peak in late February tend to carry that energy into March because they haven't been ground down by the Big 12/SEC/Big Ten gauntlet. Fresh legs matter in a tournament where you might play four games in eight days.

The biggest edge from this data isn't picking the champion. It's knowing which popular picks are built on name recognition rather than recent results. Houston as a 2-seed looks safe on paper. The BXS trend says otherwise. That's the kind of disagreement between reputation and data where brackets are won and lost.