2026 March Madness Sweet 16 Picks & Predictions

Thursday & Friday, March 26-27

The Sweet 16 is here, and we’re shifting gears on the model. Through the first two rounds, our projections used full-season efficiency data — and a quick note on that: when we ran the full-season model against the Sweet 16 lines, the outputs landed almost exactly on the posted spreads for every game. That’s a meaningful signal. It tells you the oddsmakers are building these lines from the same foundation — season-long efficiency and tempo data. When your model matches the market dead-on, the market is probably right.

So where does the edge come from? The last seven games. For the Sweet 16 and beyond, we’ve shifted to a rolling recent-form model that weights only the last seven games for each team. Recent form captures momentum, injuries, rotational adjustments, and the tactical evolution that happens as teams heat up or cool down heading into March. Predictably, some of those outputs diverge from the posted lines — in a few cases, quite sharply. Those divergences are the story of this preview.

One additional note before the breakdown: Houston hosts their Sweet 16 game with massive home crowd support, playing at their de facto home venue. That’s a real factor emotionally, but it’s worth acknowledging that the Cougars play only a game or two per season at this building, so there’s no genuine court-familiarity edge to price in. The crowd will show up. The home-court advantage in the traditional sense will not.

Thursday & Friday Breakdowns

Houston -3.5 vs. Illinois | Total: 140.5 Recent-form model: Houston 83, Illinois 70

The model strongly agrees with the market’s directional read here, but not its magnitude. Houston wins by 13 in the recent-form projection, covering the 3.5-point spread with room to spare. The crowd factor is real — this is as close to a home game as the Cougars will get in this tournament — and Illinois, despite its elite offensive efficiency (1.232 season-long), has been inconsistent in the back end of their schedule. Recent-form data appears to be pulling the Illini down meaningfully. The projected total of 153 against a 140.5 line is a 12.5-point over gap — one of the larger over signals we’ve seen all tournament. The combination of Houston covering and a notable over lean makes this the most straightforward game on the board from the model’s perspective.

Nebraska -1.5 vs. Iowa | Total: 133.5 Recent-form model: Iowa 69, Nebraska 67

Iowa wins outright according to the recent-form model, despite Nebraska being listed as a 1.5-point favorite. Nebraska has never won an NCAA Tournament game in program history — a fact that became relevant when they survived Troy in the first round — and the Cornhuskers barely advanced past Vanderbilt in the second round in what the model had projected as a virtual coin flip. Iowa, meanwhile, has been playing their best basketball of the tournament. The recent-form model sees the Hawkeyes as the slightly better team right now. Model total of 136 is a mild over lean against 133.5 — nothing actionable. The result call is the story: Iowa to upset Nebraska and reach the Elite Eight.

Purdue -7 vs. Texas | Total: 148.5 Recent-form model: Purdue 89, Texas 76

Purdue covers comfortably and the recent-form model shows the most dramatic over lean of the entire Sweet 16 slate. A projected combined score of 165 against a 148.5 line is a 16.5-point gap that’s hard to ignore. Braden Smith — now the all-time Division I assists leader — has been playing at an extraordinary level over the last seven games, and Purdue’s offensive efficiency (1.224) leads all eight Sweet 16 teams. Texas, meanwhile, has a defensive efficiency of 1.080 that has been getting carved up by quality offenses recently. The model sees this as a track meet that gets away from the Longhorns in the second half. Purdue covers, and the over at 148.5 may be the strongest totals signal of the Sweet 16 weekend.

Arizona -8 vs. Arkansas | Total: 167.5 Recent-form model: Arizona 99, Arkansas 82

Arizona wins by 17 in the model — easily covering the 8-point spread — and the projected total of 181 sits 13.5 points above the posted 167.5. Both signals are consistent: the Wildcats have been the most complete team in the tournament and the recent-form data reflects that, while Arkansas’s defensive liability (1.068 season DE, and apparently worse recently) is getting fully exposed. This is a high-octane West Region clash with significant pace (72.6 projected possessions), and the model sees it running up the scoreboard. Arizona advances convincingly, and the over at 167.5 appears underpriced.

Duke -7 vs. St. John’s | Total: 142 Recent-form model: Duke 69, St. John’s 68

The model’s most intriguing output of the Sweet 16. Duke wins by a single point — nowhere close to covering the 7-point spread. The full-season model had this lined up exactly with the posted spread, which means the divergence is entirely recent-form driven. St. John’s has been on a run, Rick Pitino’s team plays with an intensity and defensive identity that translates well in the tournament, and the Blue Devils have shown vulnerability in close games this season. Duke wins, but this is a cover alert against the Blue Devils at -7. The projected total of 137 sits 5 points below the 142 line — a modest under lean that is the only qualifying under signal of the Sweet 16 slate so far. Duke wins it, St. John’s covers, and the under has some merit.

UConn -1.5 vs. Michigan State | Total: 136.5 Recent-form model: Michigan State 81, UConn 70

Michigan State wins outright — the biggest recent-form upset call of the Sweet 16. UConn is listed as a 1.5-point favorite, but the model has the Spartans winning by 11. Michigan State’s recent-form numbers have clearly been trending upward, while UConn shot just 20% from three against Furman before Tarris Reed Jr. single-handedly kept them alive. The Huskies’ offense has looked inconsistent when Reed isn’t dominant. Dan Hurley’s teams have historically elevated in March, but the last-seven-games framework suggests MSU is the hotter team right now. This is a direct contradiction of the market — a team listed as a slight underdog projected to win by double digits on recent form. The projected total of 151 against a 136.5 line is a 14.5-point over gap, the second-largest of the weekend. Michigan State straight up, and the over.

Michigan -10 vs. Alabama | Total: 175 Recent-form model: Michigan 91, Alabama 85

Michigan wins in the model — the direction is right — but the margin of 6 points falls well short of the 10-point spread. Alabama covers per the model’s projection. The Crimson Tide’s defensive efficiency (1.081) has been an issue all season, but their offense (1.188) keeps them in games, and the recent-form model sees them playing competitive basketball down the stretch. The projected total of 176 is essentially dead-on to the 175 line — no totals edge here, just a straightforward spread signal. Michigan wins, Alabama covers the 10.

Iowa State -4.5 vs. Tennessee | Total: 138.5 Recent-form model: Iowa State 78, Tennessee 68

Iowa State covers by 10, and the projected total of 146 represents a 7.5-point over gap against the 138.5 line. The Cyclones have been dominant in their recent form, with the best defensive efficiency of any team still standing in the Midwest Region (0.934). Tennessee, despite their solid profile, has struggled to generate offense (1.137 OE, the lowest of any Sweet 16 team) and the model sees that limitation being exposed over four quarters against Iowa State’s defense. The Cyclones are a legitimate Final Four program, and both signals — cover and over — point in the same direction here.

Sweet 16 Summary

Model upsets (outright): Iowa over Nebraska, Michigan State over UConn

Non-covers despite wins: Duke (-7, wins by 1), Michigan (-10, wins by 6)

Strongest totals signals: Purdue/Texas (+16.5 over), Michigan State/UConn (+14.5 over), Arizona/Arkansas (+13.5 over), Houston/Illinois (+12.5 over)

Only under lean: Duke/St. John’s (-5 below line)

The biggest takeaway from the recent-form model is how much the landscape has shifted from the opening weekend. Where full-season data validated the posted lines almost exactly, the last-seven-games framework is projecting meaningful divergences across six of the eight matchups. Michigan State over UConn and Iowa over Nebraska are the two outright upset calls to watch. The totals landscape is almost uniformly pointing over — the market appears to have set Sweet 16 totals conservatively, and recent scoring trends suggest several of these games could blow past their numbers.

Lines as of March 24, 2026. All lines subject to movement.