LEDGER
Every week, both models' spreads next to the betting line, as they were published before kickoff — then scored against the result. Both models have been tested against the closing line and neither adds anything it does not already know; this page is that comparison, kept honestly, in public.
| Game | Line | Team Elo | Factor | Result | Closest |
|---|---|---|---|---|---|
 SEAat DEN10-15 | SEA −3.0 (6.0) | SEA −1.7 (4.7) | SEA −1.3 (4.3) | SEA 14–DEN 17 (DEN +3) | Factor |
 ARIat LA10-18 | LA −13.5 | LA −11.8 | LA −15.4 | 86% LA | — |
 BALat CLE10-18 | BAL −4.5 | BAL −5.3 | BAL −8.9 | 33% CLE | — |
 BUFat LV10-18 | BUF −6.0 | BUF −5.3 | BUF −8.5 | 33% LV | — |
 CARat PHI10-18 | PHI −1.5 | PHI −2.9 | PHI −5.0 | 62% PHI | — |
 CHIat ATL10-18 | CHI −3.0 | CHI −1.0 | CHI −1.0 | 48% ATL | — |
 DALat GB10-18 | GB −1.5 | GB −1.6 | GB −2.8 | 57% GB | — |
 HOUvs JAX10-18 | JAX −3.0 | JAX −2.3 | JAX −2.5 | 58% JAX | — |
 LACat KC10-18 | KC −8.5 | KC −5.4 | KC −6.5 | 70% KC | — |
 NOat NYG10-18 | NYG −1.5 | NYG −3.2 | NYG −2.7 | 63% NYG | — |
 NYJat NE10-18 | NE −7.5 | NE −9.2 | NE −14.9 | 80% NE | — |
 PITat TB10-18 | PIT −3.0 | PK | TB −2.1 | 51% TB | — |
 TENat IND10-18 | IND −7.0 | IND −6.1 | IND −11.3 | 72% IND | — |
 WASat SF10-19 | SF −7.5 | SF −8.8 | SF −12.2 | 79% SF | — |
Spreads are shown as the favorite and the number, home-oriented in the data. After a game the bracketed figure is each spread's distance from the actual margin; the nearest is bold.
| Wk | n | Line | Team Elo | Factor |
|---|---|---|---|---|
| 1 | 16 | 11.6 | 11.9 | 12.3 |
| 2 | 16 | 11.6 | 11.7 | 11.8 |
| 3 | 16 | 8.3 | 8.3 | 9.5 |
| 4 | 16 | 7.3 | 6.4 | 6.1 |
| 5 | 15 | 10.4 | 9.7 | 9.7 |
| 6 | 1 | 6.0 | 4.7 | 4.3 |
Reference: over 2018–25 regular seasons the closing line missed the margin by 9.86 points a game and the Factor Elo model by 10.27(closing line vs the possession model, n=2,119). A season of NFL games cannot separate two spreads that close — read the tally as a running comparison.
Team Elo: the board's walk-forward rating; the spread is the rating gap plus a home edge, at 15.6 rating points per point of margin — a derived scale, never re-fitted to the market.
Factor: the Factor Elo model's spread — points per possession from offense, defense and the announced starting quarterback, with a team-phase penalty when the starter is not the established one. Home advantage restored from the model's own measured per-possession edge.
Line: nflverse's spread for the game — the closing line once the game is played, the current line before it. Positive in the data means the home team is favored; shown here as the favorite and a negative number, the way a book prints it.