Audit the dashboard

Methodology

Formulas, score and record rules, qualifications, coverage gaps, and source boundaries for the season selected by the data manifest.

Situational rates use completed REG games. Eligible plays require canonical posteam, nflverse pass/run type, and valid success and yards; deleted/aborted plays, kneels, and spikes are excluded, while sacks classified as passes remain passes. Success rate is successful/eligible plays. Explosives are passes ≥20 yards and rushes ≥10. Third down uses converted/(converted + failed) flags. A red-zone trip is a unique fixed team drive with an eligible play at yardline_100 ≤ 20; it is a TD trip when that drive has a nondeleted offensive TD with td_team=posteam. Giveaways are interceptions plus lost fumbles. Rows are emitted only as reciprocal game pairs, and any scoped play-by-play gap makes the rate unavailable, never zero.

Schema v5Manifest-selected season · completed games

Games, scores, and records

Only completed regular-season games in the selected schema v5 snapshot appear in the scoreboard. Scores and dates come from the games collection. A team receives a win when its score is higher, a loss when lower, and a tie when equal. Records are derived through the selected week from those completed games; a bye adds no game and no result.

Fact rows join to schedules by immutable game ID rather than team and week alone. Total yards in game detail combine team passing yards with joined player rushing yards. The available turnover context is interceptions, so it is labeled rather than presented as all turnovers.

Efficiency and weighting

All totals are additive before division. The dashboard never averages already-rounded rates.

Total EPA/play = (Σ pass EPA + Σ rush EPA) / (Σ available pass EPA plays + Σ available rush EPA plays)

Passing EPA/dropback = Σ pass EPA / Σ pass EPA plays

Pass EPA plays include attempts plus sacks when EPA is available. Rushing EPA/carry and receiving EPA/target divide additive EPA by their available carry and target samples.

CPOE estimate = Σ (weekly CPOE × eligible attempts) / Σ eligible attempts

CPOE is an attempt-weighted estimate. Passer rating uses the standard four-component formula with every component clamped from 0 to 2.375, then multiplied by 100/6. Defense allowed is constructed from each game opponent’s offense; lower EPA allowed is better.

Dashboard qualifications

Let G be games played through the selected week by a player’s latest team. Traded-player totals include all teams, but the latest team supplies G.

Passing is restricted to QB. Rushing is restricted to RB/FB/QB. Receiving is restricted to WR/TE/RB/FB. Defensive pressure is restricted to the standard front-seven source positions DE/DL/DT/NT/LB. Passing QBs require max(25, 14G) attempts. Rushing RB/FB require max(15, 6G) carries; QBs require max(10, 2G). Receiving WR/TE require max(15, 3G) targets; RB/FB require max(10, 2G). Pressure players require max(50, 15G) covered defensive snaps.

Ranking and ties

Ranks are computed globally before team or name filters. Unrounded values determine order. Exact ties receive competition ranks, displayed as T-N; the next rank skips occupied places. Deterministic display ties break by sample descending, English base-name collation, then stable player or team ID. Null is always last. Unqualified players receive no rank.

Snap context

Player detail uses the role-relevant side of the ball: offense snaps and offense percentage for passing, rushing, and receiving; defense snaps and defense percentage for pressure. Weekly rows join by player and game ID. Percentages are source-reported participation rates, while counts are additive. Scoreboard snap context reports roster coverage and team-high percentages, not team play totals.

Defensive pressure proxy

Pressure rate = Σ PFR pressure credits / Σ covered defensive snaps

Pressure share = Σ player pressure credits / Σ additive team player pressure credits

Weighted disruption index = (Σ sacks + .75×Σ hits + .75×Σ hurries) / Σ covered defensive snaps

This is a public PFR-derived pressure-credit proxy, not true pass-rush efficiency. Credits and components can overlap by source convention. PRWR is shown only when the selected snapshot provides verified wins and opportunities; no other metric is relabeled as PRWR.

Known coverage gaps

Coverage is read from each selected season’s sourceAvailability and missingTeamWeeks metadata. Missing advanced team-games are displayed as unavailable, never zero. Player efficiency and snap counts have separate source boundaries and denominators.

Provenance and caveats

The frontend first requests /data/manifest.v1.json, selects its schema v5 default season, and loads the declared snapshot. A direct schema v5 path is used only when the manifest is absent in development. Source URLs, retrieval timestamps, row counts, framed input SHA-256, availability flags, and coverage counters are embedded in each snapshot.

EPA and CPOE inherit nflverse model and play-classification assumptions. PFR-derived pressure credits inherit public charting definitions. Team aliases are normalized by the pipeline. Headshots are allowlisted NFL or positive-ID ESPN assets; team marks use a fixed ESPN map with local code/color fallbacks.