The excuses now have receipts.

The rules of the argument

1. One completed-week boundary

Only final, validated weeks at or below the published release cutoff enter calculations. Rosters beyond the published cutoff are retained separately; their results, injuries and outlook do not enter this release.

2. Actual scores are Yahoo's scores

All starter and bench totals reconcile to the imported pages. Stored reception rules conflict with those scores; no recalculation or silent scoring correction is applied. Short stat summaries are not full game logs.

3. Schedule-neutral expected wins

For each week: (number of other teams outscored + half of ties) / 11. Sum across weeks. Schedule swing = actual wins + half actual ties - expected wins. Opponents' players do not affect your lineup's score, but matchup pairing changes the record.

4. House Scoring WAR

Expected wins minus half the number of weeks. It measures wins above an average .500 baseline. Scoring WAR + schedule swing equals actual wins above .500. It is not literal player replacement WAR.

5. Decision process is not hindsight

The legal projected lineup and legal hindsight lineup are optimized separately across QB, two RB, two WR, TE, flex, superflex, K and DEF. No IR player, duplicate player or illegal third QB enters. Small projection differences are close calls. A material deviation is a review flag, not proof of bad managing.

6. Forecast provenance and lock times

Yahoo Proj/Orig Proj are the provisional recorded benchmark. Retrospectively displayed FP Proj/ECR, injury badges and later rankings are not reliable pre-kickoff forecasts and are excluded. A zero forecast with positive production requires review. No point-level manager penalty is assigned without valid availability and lock-time evidence.

7. Injuries: event, timing, return, knowledge

Each tag records a source, event week, category, return status and timing where known. A star's unexpected injury does not alter the pregame process label. Pregame out and in-game exit are different; Monday inactive news may arrive after Sunday alternatives locked. Missing context is unknown, not healthy.

8. Injury loss is not forecast minus actual

Forecast shortfall mixes injuries, efficiency, game script and model error. Causal points lost require a supported estimate of remaining opportunity. Until then the ledger reports sourced events and partial exposure, not fabricated injury-adjusted wins.

9. Player WAB, not a fake waiver wire

For each starter, compare all-play share at the actual team score with share after replacing that player by the mean actual output of three other teams' highest-projected, same-position bench players. Pools are selected on forecasts, not realized rank. Fewer than three means not rated. Marginal values cannot be summed into team wins.

10. Small sample and incomplete research

Three weeks are evidence, not a settled skill ranking. Injury coverage is a documented lower bound. Missing FAAB/transaction records and incomplete historical projection timestamps remain visible limitations. Published methods are versioned so later improvements do not silently rewrite the argument.

Source of league data: commissioner-provided Yahoo SAE league 81402 box scores through Week 5, with source-linked import records. Injury sources are linked on the ledger. No paid Fantasy Points analysis, private trade targets, or personal chat archive is published.