| Player ⇅ | Pos ⇅ | Matchup ⇅ | Dep ⇅ | Line ⇅ | Proj ⇅ | Model ⇅ | Mkt ⇅ | Over ⇅ | Under ⇅ | Edge ⇅ |
|---|---|---|---|---|---|---|---|---|---|---|
| Drake Maye NE @ SEA | QB | NE vs SEA | 1 | 232.5 | 258.5 | 0.572 | 0.500 | -114 | -114 | |
| Sam Darnold SEA vs NE | QB | SEA vs NE | 1 | 234.5 | 238.1 | 0.472 | 0.500 | -115 | -115 |
Pass yards is suppressed. -16.0% 2024 / -10.9% 2025 on closing prices; corr(model, market) = -0.12. The line, the odds and the model's own number are still shown so the market keeps being measured, but the edge column is deliberately blank — graded across two full seasons this market lost money while reporting an edge, and its probability ran negatively correlated with the market's.
Cold start. No player has in-season usage yet, so the volume model is running on last year and the depth chart, and the touchdown model uses a sharper concentration for that reason. The longshot section of the nasty board turns itself off this week: graded on 2024 and 2025 it loses in week 1 and makes money from week 2 on. Legs marked CHART are ones where the posted depth chart and the model disagree about who starts — in week 1 of 2026 the chart was right and the model was wrong every time.
Reading the columns. Proj is the model's projected value and Line is what is posted. Model and Mkt are the two probabilities of the over — the model's, and the market's after the vig is removed. Edge is the better of the two sides after vig, so a blank edge means no side clears it. Dep is the player's rank at his own position on the posted depth chart.
Projections only. Nothing here is advice, and prices move.