Where the model disagrees with the Week 1 betting board
Four Week 1 games where the model's spread is more than two points away from the posted line — including a five-point gap on Chargers-Cardinals.
The model posts a number for every Week 1 game before looking at the market, which makes the comparison honest. Most of the sixteen games land within a point or two of the posted line — that agreement is the boring, healthy majority. The interesting cases are the four below, where the two numbers are far enough apart that one of them is meaningfully wrong.
| Game | Book line | Model spread | Model win prob. | Gap |
|---|---|---|---|---|
| Cardinals at Chargers | LAC -10.5 | LAC -5.5 | LAC 69% | 5.0 |
| Dolphins at Raiders | LV -3.5 | MIA -1.5 | MIA 55% | 5.0 |
| Packers at Vikings | MIN -1.5 | MIN -4.5 | MIN 65% | 3.0 |
| Browns at Jaguars | JAX -7.5 | JAX -6.0 | JAX 70% | 1.5 |
Chargers -10.5 is the market pricing a blowout the ratings don't see
Los Angeles opens at 1507 and Arizona at 1396. Add the 25-point home field constant and the gap is 136 rating points, which the logistic curve turns into a 69% home win probability and a spread near 5.5. To justify a double-digit line the market has to believe the true rating gap is closer to 260 points — roughly the distance between the best team in football and a bottom-five team. Arizona was 3-14, so the direction is not in dispute; the size is.
Miami on the road is the one outright disagreement
Las Vegas closes 2025 as the lowest-rated team in the league at 1395 after scoring 241 points all season. Miami opens at 1456. Even after handing the Raiders the full home field bump, the model still lands on Miami by a point and a half — the only Week 1 game where it picks the road underdog outright rather than shading a favorite. That is a coin-flip pick, labelled Lean rather than Confident for exactly that reason.
What a gap is actually worth
A three-point disagreement on a near-even game is a much bigger claim than a three-point disagreement on a two-touchdown favorite, because probability moves fastest around 50%. Minnesota going from -1.5 to -4.5 shifts the implied win probability by roughly ten points. Jacksonville going from -7.5 to -6.0 barely moves it at all.
A gap is a hypothesis, not an edge. The edge only shows up if the gaps win more often than they lose across a full season.
Which is why every one of these is logged before kickoff and graded afterwards on the season tracker, with no quiet edits. If the model's disagreements are noise, a year of results will say so.