Conversion rates for every down and distance,
Goal to go uses offense quality only. Field position is set by the distance, and game situation did not improve predictions inside the 10.
Each square is the chance to move the chains, using the situation filters above. Click a square to look it up. Hatched squares have fewer than actual plays, so lean on the estimate there.
This play is the chance this one snap gains the yards or scores. This series is the chance the offense moves the chains before it gives the ball up, including later downs. A punt or field goal counts as a failure. On 4th down the two numbers are the same.
4th-down numbers only include teams that chose to go for it. Coaches go for it when they like their chances, so these rates probably run higher than an average team would manage in an average spot.
Filters split the data thin, so the estimate comes from a model fit to every snap at that down. It combines yards to go, field position, game situation and offense quality. A filter you choose is applied to every snap at that down and distance. A filter left on All keeps the real mix. Each filter was kept only if it made predictions better on 2023 to 2025 games the model never saw.
The big number shows the actual rate when there are at least plays, and the estimate otherwise.