ETHICALFANTASY

We measure the difference between predicted rank and actual finish. QB1 finishing QB12 is an 11-place error. Lower average error is better.

Each position uses the same cutoff for everyone: QB 24, RB 48, WR 48, TE 24, K 20, and DEF 20. Predictions and finishes beyond that cutoff are capped at cutoff + 1. An unranked receiver who finishes WR1 receives a 48-place error (49 − 1).

Everyone is measured against the same player pool: anyone in any eligible analyst’s top cutoff, plus actual top finishers. A missing or incomplete list is excluded, never treated as a prediction. Tied fantasy points share the average finishing rank.

Position scores normalize errors to 0–100, with higher better. Each position and validated week gets equal weight; coverage accompanies the score. An overall score requires all six positions for each included week. Flex is available for browsing but excluded from the combined score to avoid counting players twice.

The displayed accuracy score is field adjusted after the selected scoring format, position, and season or week are applied. A score of 50 is the field average, and every 10 points represents one standard deviation from that average. The 0–100 scale is capped only at its endpoints. This transformation preserves the underlying leaderboard order while making real performance gaps easier to see. The leaderboard shows the top 50.

Updates are welcome before each player’s kickoff. Captures remain versioned, and a player’s last eligible pregame prediction is locked at kickoff. Results must be final. Historical captures without timing evidence remain available for browsing but cannot yet contribute to the leaderboard.