Every skill-position player's 2026 season in counting stats: passing, rushing and receiving yards, touchdowns, catches, carries, attempts. Not fantasy points. The numbers you would read off a box score in January.
One thing separates these from most projection sets: nobody here is assumed to play seventeen games. Every total is a per-game rate multiplied by that player's own projected availability, which comes from a durability model fitted on 2,215 player-seasons. The effect is not cosmetic. Christian McCaffrey is projected for eight more scrimmage yards a game than Jonathan Taylor and finishes 172 yards behind him, because he is projected for 11.4 games and Taylor for 13.9. A seventeen-game projection cannot show you that, and it is usually the difference between a good pick and a bad one.
Each stat starts as a per-game rate. For a receiver that means a target share applied to his team's projected pass volume, a catch rate applied to those targets, and yards per reception applied to the catches. For a back it is carries out of the team's projected rushing attempts and yards per carry on top. Touchdowns are allocated from a team total rather than projected player by player, so a team's rushing scores add up to something its offence can actually produce.
Those rates are then checked against de-vigged season-long betting markets. Where our number and the market disagree by more than 10 percent, the market wins and we move. It is real money on a live line, and it prices news faster than any model.
Projected games is the column nobody else publishes, and the one doing the most work. The durability model separates chronic absence, the recurring soft-tissue kind that predicts itself, from catastrophic injury, which mostly does not. It is fitted to the median season rather than the mean, because fantasy drafts the typical year and you can replace a hurt player weekly.
Totals are rounded for display: yards and attempts to whole numbers, touchdowns to a tenth. A back projected for 9.4 rushing touchdowns is not going to score four-tenths of anything. That is an expectation across the season, and rounding it to 9 would throw away real information about the gap between him and a back at 8.6.
| Passing | Rushing | Receiving | Total | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Player | Pos | Tm | GP | Cmp | Att | Yds | TD | Int | Att | Yds | TD | Tgt | Rec | Yds | TD | Scrim | TD | PPR |
Every stat begins as a per-game rate built from the player's usage and his team's projected volume: target share, carry share, catch rate, yards per touch. Those rates get checked against de-vigged season-long betting markets, and where we disagree with the market by more than 10 percent, the market wins. The rate is then multiplied by his projected games to produce the season total in the table.
No. That is the main thing setting them apart. Most published projections quote a full seventeen games and leave you to apply the injury discount in your head, which nobody does consistently. Here the discount is already in the number, taken from a durability model fitted on 2,215 player-seasons that treats chronic absence and catastrophic injury as the different things they are. Only a handful of players carry a full sixteen-game projection.
Tuesday mornings once the previous week has settled. In the preseason they move whenever the news warrants it: a depth-chart change, a trade, an injury that moves someone's expected games. The date above the table is the honest one.
Looking for something narrower? The rest-of-season rankings order the same projections by value over replacement for your league's scoring, best ball ranks them on what each player adds above your own bench, and dynasty values them across eight seasons instead of one.