CORE
Contextual On-Field Rating Estimate
expected points per 100 qualifying snaps
Independent NFL analytics · completed 2025
See the game underneath the game.
Twenty-two seasons of games and team context, plus published 2024–25 player opportunity — with the evidence one click away. Real football only: not fantasy, not betting.
completed 2025 · no 2026 performance claims 22 seasons of published football
Where the season is
What this edition can answer right now
Four separate clocks, because they are four different questions: when football was last played, when the sources were last observed, when the outputs were built, and when the edition was published.
Enable JavaScript for the live publication state, or read season-state.json.
2025 at a glance
Four answers from the completed season
Ranks and bars are descriptive presentation of source-attributed EPA. Player ratings remain blocked until they earn promotion. They are not a power ranking and not a prediction.
The league map
Where every offense met every defense
Better offense sits right. Better defense sits higher. Dividing lines are the 2025 league means. Every point opens that team.
Rank the league
Who drove the league
Accuracy against impact, for every qualified quarterback
All 38 quarterbacks with at least 200 dropbacks in 2025, placed by completion percentage over expectation against expected points added per dropback. Descriptive, source-attributed facts — not a rating; none are published.
Enable JavaScript, or read the published quarterback comparison.
Games that swung
The season finale, drive by drive
The last game of the completed 2025 season, reconstructed from the published record: every possession, what it ended in, and the play that swung the game hardest.
Enable JavaScript, or browse every 2025 game.
Who got the work
Who saw the ball in 2025
Opportunity counts, each under the name of the exact metric it measures. Volume facts, not rankings of quality — no player rating exists on this platform yet.
The last slate
Recent games
How we know
What the data can and cannot prove
The latest completed season is 2025. Games, teams, and player opportunity are live. Player ratings are not published yet. The current NFL season is 2026; this preview makes no 2026 performance claims.
Publication state
Generated from the committed manifest, from 104 source bindings in that frozen evaluation manifest. from the committed publication manifest
104 source bindings in the frozen August 8 evaluation manifest. Later descriptive 2026 identity captures are outside that completed-season evaluation count.
22 of 27 seasons (22 of 27) pass the publication gate; the rest stay visibly quarantined rather than quietly dropped.
No player ratings are published yet. 2025 games, team context, and who actually saw the ball are here. A player-impact number appears only when one has earned a public value — see Methodology.
- Season: the latest completed season is 2025, and the current NFL season is 2026. Nothing here is a 2026 performance claim.
- Coverage: twenty-two seasons of published football are open. Five older seasons stay quarantined because the source record is incomplete — they remain visible rather than quietly dropped (which seasons, and why).
- Rights: share-alike inputs stay on their own lineage; proprietary comparators never enter what this site publishes.
Twenty-seven seasons, and the seams between them
Enable JavaScript for detail, or read coverage.json.
The research ledger
What we tested lately
Not published yet
The metric family
CORE, FORGE, PULSE, and SIGNAL are the rating families. None of them publishes a public player value yet. Methodology explains how a metric earns promotion; this page shows a number only when one has.
FORGE
Football Overall Replacement-adjusted Game Equity
expected points, then wins when calibrated
PULSE
Performance Under Latest Sample Evidence
window-dependent rate
SIGNAL
Sample · Identifiability · Granularity · Noise · Availability · Lineage
six axis scores, 0–100
The hard problems
The trench moonshot
Offensive line, interior defensive line, off-ball linebackers, and coverage defenders who rarely see a target are where public football analytics either goes silent or goes dishonest. Gridiron Signal treats them as flagship research problems: competing model families, identifiability scores, unit-versus-player attribution, and published disagreement between models — wider uncertainty instead of cowardly omission, honest intervals instead of fake precision.
Positions publish in evidence order: quarterbacks first, then backs, receivers, tight ends, specialists, edge rushers and targeted defensive backs — and the trenches as the boss battle, attempted seriously.