Yards per carry

What it measures

The average length of a player's runs. That sounds like rushing efficiency, but the distribution underneath is the problem: most carries gain 2 to 4 yards, and the average is dragged around by a handful of breakaways. Two backs can post identical 4.6 marks where one grinds out steady first downs and the other alternates stuffs with a monthly 60 yarder. Same number, completely different players.

How it is computed here

Rushing yards divided by carries, where carries count rush attempts including quarterback scrambles and kneels and exclude two-point tries. We ship the raw carries and rushing_yards columns alongside the ratio so you can always check the volume behind the rate, plus epa_per_carry when you want value instead of distance. Kneels counting as carries is worth remembering: a quarterback's own YPC is polluted by clock-killing.

How fast it stabilizes

Barely at all, and this is the headline. Because a single long run can move a season's YPC by half a yard, the stat carries little repeatable signal even across full seasons; a back's YPC one year tells you surprisingly little about the next. Below roughly 100 carries it is close to meaningless. Success rate on carries and EPA per carry both stabilize faster because they cap or contextualize the influence of outlier runs.

How it gets misused

• "He averaged 5.0, feed him more." The average is a breakaway artifact more often than a repeatable rate, and efficiency tends to fall with workload anyway.
• Grading run blocking or backs in isolation. Box counts, game script, and scheme all live inside YPC, and it cannot separate them.
• Comparing quarterbacks to running backs. Scrambles are opportunistic runs against retreating coverage; designed carries are not the same population.