How the rankings work
The rankings score what has already happened this season. This page covers how a player's composite score is put together, what each setting means, and how well the rankings hold up in testing.
Summary
- Each player is measured on several statistics, and each one is turned into a score out of 100 by comparing him with other players at the same position.
- Those scores are grouped into how efficient he is (how much he does per chance he gets) and how much he has produced (raw totals), then blended into one number.
- This describes the season so far. It is not a prediction; for that, see how the predictions work.
Composite score
Every player is measured on several statistics. Each one becomes a percentile among ranked players at the same position that week, so a 90 means better than 90% of them. The percentiles are averaged into two scores from 0 to 100, then blended:
- Efficiency: rates per opportunity, such as EPA per dropback or yards per target.
- Production: totals and shares, such as yards, touchdowns, and target share.
Fantasy points
The second ranking is season-to-date PPR fantasy points (one point per reception). The data source does not score kickers, so I use common standard scoring for them.
Things to keep in mind
- Early-season rankings are noisy. In the first few weeks every player has only a handful of plays behind them.
- Only players with enough volume are ranked. The minimum grows with each game the team plays, so a few good plays never rank a player high late in the season. Everyone else is listed separately.
- Rankings only use games already played. Each week's ranking uses data through that week and nothing later.
- These are statistical rankings, not predictions or betting advice.