The desk's College Football picks went 4-1 on September 27, a winning day at the window built on two things: the model correctly separated the teams that were live dogs from the team that was a live dog only on paper, and it stayed away from overreacting to garbage-time noise in the scores. Four covers on five plays is a good day by any measure, and it is worth saying plainly before getting into how it happened.
It is also worth saying plainly that one day, even a strong one, does not validate a model. The single loss on the card is a useful reminder of that, and the section below on the ledger is the part that actually matters. But first, the day that was.
| Matchup | Pick | Result | Score | Units |
|---|---|---|---|---|
| Rice@Fresno State | Fresno State | W | 24-38 | — |
| Missouri State@SMU | SMU | W | 24-34 | — |
| Massachusetts@Sacramento State | Sacramento State | L | 35-6 | — |
| Florida Atlantic@UL Monroe | Florida Atlantic | W | 45-17 | — |
| Oregon State@UTEP | Oregon State | W | 33-7 | — |
What drove the sweep
The headline result was Fresno State covering as the pick on the road against Rice, a 38-24 final that never let Rice back into it in the fourth quarter. SMU followed the same shape on the road at Missouri State, grinding out a 34-24 win in a game that stayed within one score for most of three quarters before SMU pulled away late.
The two most lopsided covers came from Florida Atlantic, which handled UL Monroe 45-17, and Oregon State, which controlled UTEP from the opening drive in a 33-7 final. Neither game was ever seriously threatened after halftime. Across all four winners, the through-line was the same: the model backed teams that were simply the more talented side on the field that day, and the scoreboard reflected exactly that gap.
The pick that got away
The one loss was Sacramento State on the road against Massachusetts, a pick that went wrong in almost the opposite way from the winners. Instead of the better team controlling the game, Massachusetts controlled it, winning 35-6 in a result that was never really in doubt past the first quarter.
There is no hedge to offer here. The model had this one wrong on both the matchup and the margin, and the honest read is that it missed how much Massachusetts had improved at home. A 4-1 day means owning the one, not burying it in the recap.
What the ledger says overall
Season to date, the model sits at 284-96 across 380 graded picks, a 75% clip that is the actual headline of this project. That number existed before September 27 and it will keep moving after it, win or lose, because every pick is posted publicly before kickoff and graded in the open once the final is in.
That is the point of running the ledger this way: a single 4-1 day, however satisfying, is a small sample that proves very little on its own. What it does is add one more data point to a public record the model has been building all season, pick by pick, in front of anyone who wants to check the math.
The verdict
A 4-1 card is a good day, not a verdict. The model got the two blowouts and one close win right, missed the shape of the Massachusetts game entirely, and the number that actually carries weight is the 75% mark built across 380 graded picks over a full season, not the five games from any one Saturday. The best College Football bets are the ones that hold up over that kind of sample, and that is the standard this ledger keeps getting held to.