Every model on this site is a claim, and a claim you can’t check is a black box. Research is where the claims get tested: what actually carries year to year, where touchdown luck comes back, which of the things scouts measure translate to snaps. Each study ships with the chart, the method and the data behind it — and a builder so you can rerun the question with your own axes.
Any metric against any other, for any position or team, across four seasons. Start from a preset, then change an axis. Export the result.
How a player is used, not how good he is. Usage-pattern clusters — alpha, deep threat, possession slot, satellite back — stabilized across seasons.
Week-to-week distribution rather than the season total. Floor, ceiling, boom rate, bust rate and the volatility that a per-game average hides.
Which production was earned and which was borrowed. Efficiency residuals, touchdown luck and the players the market has not repriced yet.
Define a group by any set of filters — rookie WRs over 200 lb, backs with 250+ touches — and watch what that cohort did next season.
Year-over-year self-correlation for every metric in the catalog. The honest answer to “is this stat real or is it noise?”
Every study states its question, its sample and its cut-offs before it shows a number, and links the exact Chart Builder view that produced the figure — so the first thing you can do with a conclusion is try to break it. Where a result is thin, small-sample or vintage-sensitive, it says so in the study rather than in a footnote nobody reads. Definitions live in the Glossary; model construction lives in the study itself.