Week 3
Model against market.
The model's home win probability beside the market's, ordered by where the two disagree most. Probability is the common unit: comparing one moneyline against another is not linear and reads as nonsense, so both sides are converted and the prices are shown alongside rather than compared directly.
The market's margin is removed first. A posted two-way price sums to more than certainty, about 4.2 percent more this week, because that difference is the book's. Comparing a clean model number against an inflated market one would overstate the disagreement on every favourite.
Lines recorded Sun Sep 27, 12:00 PM ET. live-20260923T042310Z.
The largest disagreement this week is KC at MIA: the model gives MIA 46.3% where the market implies 17.8%, a gap of +28.6 points.
| Game | Kickoff | Model | Market | Gap | Fair ML | Market ML | Spread | Total |
|---|---|---|---|---|---|---|---|---|
| KC at MIA | Sun Sep 27 1:00 PM | 46.3% | 17.8% | +28.6 | -116 / +116 | -600 / +440 | -1.2 / -10 | 45.9 / 45.5 |
| CIN at PIT | Sun Sep 27 1:00 PM | 58.6% | 38.4% | +20.2 | +141 / -141 | -180 / +150 | +2.9 / -3.5 | 43.9 / 42.5 |
| PHI at CHI | Mon Sep 28 8:15 PM | 56.1% | 36.3% | +19.8 | +128 / -128 | -198 / +164 | +2 / -3.5 | 45.8 / 42.5 |
| SEA at WAS | Sun Sep 27 1:00 PM | 39.8% | 22.8% | +17.0 | -151 / +151 | -410 / +320 | -3.3 / -8.5 | 43.6 / 40.5 |
| ARI at SF | Sun Sep 27 4:05 PM | 64.7% | 77.2% | -12.5 | +183 / -183 | +320 / -410 | +5.2 / +7.5 | 46.9 / 48.5 |
| BAL at DAL | Sun Sep 27 4:25 PM | 53.0% | 40.7% | +12.4 | +113 / -113 | -162 / +136 | +0.9 / -3 | 48.5 / 53.5 |
| LAC at BUF | Sun Sep 27 1:00 PM | 64.2% | 74.1% | -9.9 | +179 / -179 | +270 / -340 | +5 / +7 | 45.4 / 50.5 |
| TEN at NYG | Sun Sep 27 1:00 PM | 63.4% | 54.3% | +9.2 | +174 / -174 | +110 / -130 | +4.7 / +2.5 | 43.7 / 37.5 |
| CAR at CLE | Sun Sep 27 1:00 PM | 53.3% | 45.7% | +7.5 | +114 / -114 | -130 / +110 | +1 / -2.5 | 42 / 41.5 |
| HOU at IND | Sun Sep 27 1:00 PM | 54.1% | 46.8% | +7.3 | +118 / -118 | -125 / +105 | +1.3 / -1.5 | 46.8 / 42.5 |
| MIN at TB | Sun Sep 27 4:05 PM | 54.1% | 48.9% | +5.1 | +118 / -118 | -115 / -105 | +1.3 / -1.5 | 42 / 42.5 |
| ATL at GB | Thu Sep 24 8:15 PM | 61.9% | 66.9% | -5.0 | +163 / -163 | +190 / -230 | +4.1 / +4.5 | 45 / 43.5 |
| NYJ at DET | Sun Sep 27 1:00 PM | 67.5% | 72.2% | -4.7 | +208 / -208 | +245 / -305 | +6.3 / +6.5 | 48.3 / 48.5 |
| NE at JAX | Sun Sep 27 1:00 PM | 55.0% | 59.3% | -4.4 | +122 / -122 | +136 / -162 | +1.6 / +3 | 43.9 / 45.5 |
| LA at DEN | Sun Sep 27 8:20 PM | 49.0% | 45.7% | +3.3 | -104 / +104 | -130 / +110 | -0.4 / -1.5 | 42.5 / 43.5 |
| LV at NO | Sun Sep 27 4:25 PM | 60.8% | 63.7% | -2.9 | +155 / -155 | +164 / -198 | +3.7 / +3.5 | 44.3 / 43.5 |
Model and Market are the home team's win probability. Gap is the model minus the market, in probability points, so a positive number means the model likes the home team more than the market does. The paired columns read away then home, and model then market.
A gap is information, not instruction, and it is as likely to be the model being wrong as the market. The two are not measuring quite the same thing either: a market price carries what people are willing to stake, and a model carries only what its inputs imply.