About

Play-by-play, rosters, schedules, snap counts, depth charts and draft records come from nflverse, which publishes cleaned NFL data openly. Current injuries come from CBS Sports, because the official weekly reports do not name season-ending injuries: a player who tears a ligament goes on injured reserve and stops appearing on them altogether.

Between the twenties almost every offence looks competent. Inside the 20 the field compresses, explosive plays stop existing, and teams that move the ball beautifully still fail to score. That gap is the site’s premise. Trips are counted per drive rather than per play, and a drive counts as converted if any red-zone snap on it reached the end zone.

Volume, not points. Targets per game repeat from one season to the next at r = 0.84 while yards per target manage 0.44 and touchdowns per target 0.15, so a player’s own volume is trusted, his efficiency is pulled toward the positional average by how much it actually persists, and the result is scaled by what his team projects to throw and run. Season totals are that rate times the games he is expected to play rather than a flat seventeen, which almost nobody reaches. That alone cuts the error on a season projection by nearly a third.

The rankings are then blended with the FantasyPros consensus, lightly at the top of the board and more heavily in the deep tiers where the model is weakest. The Model column shows what the model said on its own, so the two can always be taken apart. Quarterbacks are the position predicted worst, their scoring depends on an offensive line and a play-caller that nothing here measures. Strength of schedule is shown but not used: the NFL schedules a near-identical slate by design, and across three seasons it correlates with beating expectation at 0.04.