Loading
Loading
Loading
Loading
A living notebook by Riel St. Amand
The reproducible trail for RB earned yards. The efficiency-noise baseline re-runs live on player_season_pairs; the earned construction runs on rb_altitude_week.
Every cell re-runs in your browser against current data. Covers 1999 through 2025. Data refreshed 1 day ago.
We asked: how much of a running back stat line does the back actually own?
The baseline first. Yards per carry in consecutive seasons, 80 or more carries and 8 games in both years, from the public player_season_pairs view. The cell below runs it live; our published run came back at 0.18.
Rigor check
Yards per carry against Next yards per carry
Pearson correlation, re-run in your browser on current data.
What survives is the role. Earned scrimmage yards price every carry at the league rate for its box count and every target at the league rate for its depth, then ignore the outcome. The cell below shows how tightly that opportunity measure tracks delivered scrimmage yards across every back week since 2016, from the public rb_altitude_week view.
Rigor check
Earned yards against Scrimmage yards
Pearson correlation, re-run in your browser on current data.
The split-half receipts are scripted: weekly reliability .307 for earned against .201 for the box score, the .641 against .631 next-season prediction where throwing away outcomes beat the outcomes at their own game, and the unmeasurable scouting staples (stacked-box yards per carry at .005, fumble rate at .003) all use the odd-even harness described in the article backtest block. The columns they need ship in these views.
Built on data from the nflverse project.