Players — 2024 Week 5

Some seasons ahead are not ranked for RB here, because what we have for them doesn't hold up; the seasons offered above are the ones we publish.

Early read: these figures are projected for him individually. His rushing touchdowns, catches and receiving yards were each moved up and widened by as much as we measured full-time starters' seasons have actually landed above our projections for them, and outside their ranges. The range around them moves every stat he is scored on together, the way full-time starters' seasons actually move, and allows for the games he may miss at the rate full-time starters at his position have missed them; the chance he plays at all is the historical rate for players who carried his workload at his age. During the season, his figures are also adjusted for his age, by as much as we have measured full-time starters at his position decline, or improve, with age from one season to the next.

He was not a full-time starter last season. These figures start from his own most recent season and move every stat he is scored on by how much players at his position and age have actually changed over the same span. The range around them moves those stats together, the way real seasons did, reaches down to the seasons where a player's whole workload fell away, and was widened by exactly as much as we measured it had been too narrow, and the chance he plays is the historical rate for players like him.

13 players here are projected from the 2023 season, marked † below: no game in 2024. The number comes from the model that looks 3 seasons ahead of a 2023 season, to reach the 2026 season. From a 2023 season we publish through 2026 and no further.

Only 21 of the 78 players here were full-time starters last season projected for them individually, and what follows is about them: This range carries a correction we measured. We compared the season-ahead ranges we had published against the seasons that actually followed, found them about 12% too narrow and centered about 8% of a range's width too low, and moved them by exactly what that comparison measured — nothing was widened to make a number look better. The figure above is a prediction from the corrected model.

About this range: we checked ranges like it against the seasons that actually followed, and it held at both levels we test — 77% of real seasons landed inside the 80% range and 89% inside the 90% range, measured on 95 player-seasons. What that testing could not settle, using only the seasons completed by then (through 2023), is whether this projection is more accurate than a simple baseline. We publish it because the range holds, not because it has been shown to beat one.