Here is our William & Mary vs Duke prediction for Saturday, September 26, 2026, in College Football Week 4. The model's pick is Duke, and its own projection is Duke -29.1. Straight up, it has the Blue Devils winning comfortably.
The tier is PASS. That is an official-card PASS with no official action, so we are not backing this game. The write-up below explains how the model reached its side and what keeps us out of it. The moneyline on the pick is n/a.
Why the model is on Duke
Duke's route to winning is sustained offensive execution. Walker Eget has three starts behind him, and the model expects the offense to keep moving the ball against a Colonial-level defense. On the other side, the Blue Devils' defensive front is facing a lower level of competition than it saw earlier in the year.
The supporting number is 37. That is how many points Duke allowed across its games against Tulane, Illinois and Stanford. It backs the projection, though we will not oversell it. Final scores do not prove trench dominance, and the model does not claim they do.
The model also gives Duke no credit for Mensah. Official-source verification confirms he has left, so nothing in the projection leans on his production or availability. Whatever Duke does Saturday, it does without him.
The case for William & Mary
The underdog's path is simple: control the ball and shorten the field. William & Mary wants stalled Duke possessions and good starting position. The risk notes point to Duke's reduced interior-line depth as the place to attack, because Fina's season-long absence thins that group.
There is also history. William & Mary beat Charlotte 41-24 in 2022, so an FBS upset is in the program's past. That result shows a repeatable offensive efficiency, but it is not a documented current-season A-game.
The results since then cut the other way. The FBS losses that followed came by 14, 19 and 39. Those margins give the model no current evidence of comparable strength. On the quarterback side, Tyler Hughes is the readable last-starter inference, though the date on that information is stale.
The number that matters
The number is 29.1. The model kept it with a 0.0 adjustment because no verified new scheme, personnel, weather or motivational lever supports moving it. Neither team's file includes reviewed film, so there was no reason to nudge it in either direction.
That is worth pausing on. A spread this large is a claim about the gap between two rosters, and the evidence for the gap is mostly results, not tape. Health is the other soft spot. Eget's September 16 health report and Nicholas's September 19 designation are dated evidence, not fresh clearance, so the model treats them as context and not as confirmation.
What we are doing with it
Nothing. The tier is PASS, and this is not a bet we are making. Duke is the side the model prefers to win the game, and that is a separate question from whether we put a play on the card. We are not dressing it up as anything else.
If you are looking at William & Mary vs Duke on your own, treat this as the model's read: Duke wins, the margin is large, and the supporting evidence is a few dated facts and a defensive scoring total. Check the latest injury news before kickoff, especially anything on Eget.
The verdict
Verdict: PASS. The model's pick is Duke, with a projected margin of 29.1 points. We are not backing this game, and there is no official action on it. Every play and pass is graded in public on the ledger.
- Event · College Football Week 4 · Saturday, September 26, 2026
- Model · Heismann, Botcappers' AI College Football model — how it is built and what it does not know
- Record · every pick is timestamped before the event and graded in public, win or lose
- Statistics · the model's own match/game database, current to the posting time above. Figures a reader cannot open here are internal to that database, not published third-party data.