The Film RoomWeek 2Sep 11, 2026
Week 2’s Five Best Edges: The Model Is Hunting Totals
The board found five totals worth playing. Here’s the model case, the football mechanism and the strongest argument against each one.
The Card
- Texas Tech at Oregon StateUNDER 53.5 · 1uOregon State’s one-dimensional Week 1 offense meets the best defensive profile on this card.PENDING
- Georgia State at Kennesaw StateUNDER 54.5 · 1uKennesaw State’s run-heavy, clock-control path gives the Week 1 scoring outbursts room to regress.PENDING
- No. 11 Oklahoma at MichiganOVER 43.5 · 1uOklahoma can carry the scoring load if Michigan contributes two touchdowns.PENDING
- UTSA at Texas StateUNDER 66.25 · 1uUTSA’s opener was powered by four long touchdowns and a pick-six—production that can cool quickly.PENDING
- Arkansas at No. 20 UtahOVER 55.25 · 1uUtah’s balanced explosiveness gives this over more than one route home.PENDING
The five strongest Edge plays on this week’s board all land on the total. We didn’t go looking for that theme. It’s simply where the largest model-versus-market differences showed up.
One distinction first: the biggest raw gap isn’t automatically the best bet. We ranked these five by the full case—the model gap, the historical pocket behind it and whether the football evidence gives us a believable mechanism. An Edge is a historically profitable signal. It isn’t a promise.
1. Texas Tech at Oregon State — Under 53.5
Model total: 49.14 | Edge margin: 4.36 points
Texas Tech didn’t need a perfect offensive night to beat Abilene Christian 33–10. The Red Raiders gained 471 yards, but their better Week 1 signal came on the other side of the ball: 194 yards allowed and 10 points conceded.
Will Hammond was efficient in his first game back from an ACL injury. Tech also spread 32 carries across three backs instead of asking him to turn the opener into a track meet.
Oregon State’s 33–20 loss at Houston tells a similar story from the opposite direction. The Beavers threw for 376 yards but managed 26 rushing yards. That’s a difficult way to sustain offense against a defense with Tech’s front-seven profile. Oregon State can create chunk plays through Braden Atkinson and Jesse Legree. But if it can’t stay ahead of the chains, the entire offense becomes dependent on low-percentage explosives.
The case against it? Game state. Texas Tech can do most of the scoring itself, and Oregon State already showed it can generate yards through the air. A late touchdown against backups can ruin a well-read under. Still, this is the best combination of model separation and historical support among the five. We don’t need Oregon State to disappear. We need its yardage to remain less efficient than it looks.
Sources: Texas Tech Week 1 recap · Oregon State Week 1 recap
2. Georgia State at Kennesaw State — Under 54.5
Model total: 48.90 | Edge margin: 5.60 points
The scoreboard will make this one uncomfortable. Georgia State opened with a 59–10 win. Kennesaw State won 47–0. The two offenses combined for 1,086 yards. That’s not the usual sales pitch for an under.
But Week 1 opponent quality matters. So does how Kennesaw State wants to play. The Owls ran for 297 yards, held the ball for more than 34 minutes and held West Georgia to 0-for-12 on third down. That’s a clock-control path, not necessarily a shootout path. Georgia State was just as dominant, producing 563 yards while holding North Carolina A&T to 209.
The strongest counterargument is obvious: maybe the offenses are simply ahead of the defenses, and we’re leaning too hard on regression after one week. Georgia State also has 67 newcomers, including 37 transfers, so its opener may contain more real change than a normal small-sample spike. The under case is that both Week 1 scores were built in mismatches. This game should have longer possessions, fewer short fields and fewer clean trips to the end zone.
Sources: Georgia State Week 1 recap · Kennesaw State Week 1 box score
3. No. 11 Oklahoma at Michigan — Over 43.5
Model total: 49.55 | Edge margin: 6.05 points
This is the best historical pocket of the five. It’s also the pick with the clearest football objection.
Oklahoma beat UTEP 51–0 behind 6.9 yards per play, five scores in five red-zone trips and touchdown passes of 40, 34 and 48 yards. John Mateer gave the Sooners sustained offense and explosive-play access. The defense did its job, but the over doesn’t require it to suddenly become bad. It requires Oklahoma to push Michigan into a game where 20 points may be enough from the Wolverines.
That is the entire handicap. Can Michigan contribute?
The Wolverines gained 276 yards, possessed the ball for 19:52 and went 3-for-10 on third down in a 13–12 escape against Western Michigan. Their final touchdown came on a 47-yard Hail Mary. Those aren’t hidden signs of a healthy offense. Bryce Underwood did account for 217 yards between the pass and the run, and JJ Buchanan produced 126 receiving yards. There’s at least a route to explosives if Oklahoma forces Michigan out of its preferred game script.
A 27–13 Oklahoma win loses this bet. So does 24–17. Michigan’s pace and third-down problems aren’t minor concerns. They’re why this play is third instead of first despite the six-point model gap. The over works if Oklahoma’s offense is portable and Michigan finds two touchdowns. It doesn’t if the Sooners control the game without forcing the Wolverines to answer.
Sources: Oklahoma Week 1 box score · Michigan Week 1 game center
4. UTSA at Texas State — Under 66.25
Model total: 58.38 | Edge margin: 7.87 points
Largest raw gap on the board. It’s not our top-ranked play because its historical pocket has the thinnest profit margin of the group.
UTSA scored 45 in its opener, but the path matters. The Roadrunners had five offensive plays of at least 30 yards, added a 49-yard pick-six and scored on touchdown plays of 36, 75, 62 and 53 yards. That’s real explosiveness. It’s also a scoring profile that can cool quickly when even two of those plays become ordinary gains.
Texas State’s 59–7 loss at No. 5 Texas was ugly, but the Bobcats still ran 70 plays, held the ball for more than 30 minutes and gained 349 yards. They moved the ball better than seven points suggests. They just couldn’t finish drives against a national-title-caliber defense.
The danger isn’t subtle. UTSA has demonstrated big-play speed, Texas State should score more efficiently against this defense, and last season’s meeting finished 43–36. If the rivalry turns loose, 66.25 is reachable. The model’s disagreement is large enough to keep the under in the top five. Don’t confuse it with the safest play.
Sources: UTSA Week 1 recap · Texas State Week 1 recap
5. Arkansas at No. 20 Utah — Over 55.25
Model total: 59.19 | Edge margin: 3.94 points
Utah’s opener was an avalanche: 66 points, 633 yards and 364 rushing yards against Idaho. Devon Dampier threw three touchdowns, three Utes cleared 70 rushing yards and an entirely rebuilt offensive line didn’t allow a sack. The opponent adjustment is severe, but the mechanism isn’t complicated. Utah showed balance, explosiveness and enough quarterback depth to keep scoring after the first unit left.
Arkansas was less convincing in a 31–14 win over North Alabama. This over doesn’t need the Razorbacks to trade scores all night, though. If Utah’s offense remains functional against an FBS defense, Arkansas needs to land somewhere around 17–21 points for 55.25 to become reachable.
The case against it is Utah’s defense and its ability to control the game on the ground. The Utes held Idaho to 225 yards, forced turnovers on the first two possessions and can shorten the second half if they build a lead. A comfortable 38–14 Utah win still stays under. We’re betting that Arkansas contributes enough to stop Utah from turning the final quarter into a drain-the-clock exercise.
Sources: Utah Week 1 recap · Arkansas 2026 results
The board is unusually concentrated this week: five totals, three unders, two overs. That doesn’t make the card safer. It makes the model’s argument clearer.
The work now is deciding which scoring mechanisms can survive the jump from Week 1 evidence to Week 2 competition—and which box scores were lying to us.