Built from Data/player_context.csv, which is joined from nflverse's 2026 rosters, 2023–2025 season stat lines, 2023–2025 official weekly injury reports and snap counts, and OverTheCap's live contract pages, all retrieved 2026-08-02. Games played and injury-report weeks are observed. Avail and Adj VOR are models and are marked as such.
Why this note exists
The first version of this board carried a caveat it could not answer: value over replacement does not price injury risk, every projection on the board assumed seventeen games from everybody, and nothing in the build knew whether a given player had ever finished a season. That is now measured.
The measurement
Avail is the share of his team's games a player has actually played across 2023–2025, shrunk toward his position's average by the weight of one full season. The shrinkage matters: without it a rookie who played 17 of 17 would score a perfect 1.00 on a single season of evidence, which is exactly the small-sample mistake this build warns about everywhere else. Players with no NFL history carry their position's base rate, because there is nothing to measure.
Adj VOR scales each projection by Avail ÷ the position's mean Avail, then recomputes value over replacement on the adjusted numbers. The scaling is deliberately relative: a published projection already prices the average player's injury risk, so multiplying by raw availability would count it twice. An average-durability player is untouched; only players away from their position's norm move.
What Avail does not use: the official injury reports. Those columns — weeks listed Out, weeks Questionable, the body parts — are carried in player_context.csv and shown on player pages because they tell you why a player missed time, but the number itself is built only from games played. A game missed is a game missed whatever the cause.
That last point has teeth, and the clearest case is below: a player can carry a punishing Avail for reasons that have nothing to do with his body.
The finding that contradicts the received wisdom
This build — and most fantasy writing — repeats that running back is the most injury-prone position, and uses that to discount the running-back-heavy top of the board. Among the players you would actually draft, the data does not support it.
| Pos | Players | Mean availability | Seasons missing 1+ game | Missing 4+ | Eight games or fewer |
|---|---|---|---|---|---|
| QB | 20 | 0.833 | 65% | 25% | 12% |
| RB | 41 | 0.857 | 63% | 27% | 6% |
| WR | 46 | 0.861 | 67% | 24% | 8% |
| TE | 23 | 0.830 | 70% | 32% | 6% |
Read the table, not the reputation. On none of the four measures is running back the worst position. Its mean availability (0.857) sits above tight end and quarterback and below receiver (0.861); it has the lowest rate of seasons missing at least one game of any position (63% against 67% for receivers); and its rate of near-lost seasons (6%) is the joint lowest, level with TE. The position that is consistently worst is quarterback, on 1 of the four measures — near-lost seasons.
So the honest correction is not "running backs are durable". It is that running back is not the position that misses the most time, and the board's old caveat — that a running-back-heavy top needs discounting because the position is uniquely fragile — is not what the games-played record shows. Receivers are modestly more durable; quarterbacks are markedly less.
Two things that keep this from being the whole story
**The pool is survivor-selected.** These rows are players projected to matter in 2026 who were also in the league for all three of 2023, 2024 and 2025 — a back whose career was ended by injury is not in it, and neither is one who lost his job by missing a season. The measured claim is therefore narrow: *among the players on your draft board*, running back is not the position that misses the most time. The wider claim, that the position destroys careers fastest, is not tested here and may well be true. **Availability is not injury.** `Avail` counts a missed game whatever caused it. Suspensions, holdouts and benchings all land in it, and for some players that is most of the number — see the table below.Missed time that was not medical
Players whose absences mostly never reached an official injury report. Reading their Avail as fragility would be a straightforward mistake.
| Player | Pos | Games missed, 3yr | Weeks listed Out | Missed off the report | Avail |
|---|---|---|---|---|---|
| Malik Willis | QB | 39 | 0 | 39 | 0.386 |
| Greg Dulcich | TE | 36 | 5 | 31 | 0.43 |
| Keaton Mitchell | RB | 27 | 1 | 26 | 0.559 |
| J.K. Dobbins | RB | 27 | 2 | 25 | 0.559 |
| Chris Rodriguez Jr. | RB | 23 | 1 | 22 | 0.618 |
| Rashee Rice | WR | 23 | 2 | 21 | 0.625 |
| Daniel Jones | QB | 22 | 2 | 20 | 0.636 |
| George Holani | RB | 21 | 1 | 20 | 0.529 |
| ## What the adjustment moves |
The players the adjustment helps and hurts most, by change in board rank.
Helped most — durable, and the market has not paid for it
| Player | Pos | Age | Avail | Board rank | Adjusted rank | Move | Games missed, 3yr | Weeks listed Out |
|---|---|---|---|---|---|---|---|---|
| Courtland Sutton | WR | 31 | 0.93 | 101 | 64 | +37 | 2 | 1 |
| Jared Goff | QB | 32 | 0.96 | 81 | 46 | +35 | 0 | 0 |
| Matthew Stafford | QB | 39 | 0.92 | 93 | 58 | +35 | 3 | 0 |
| Michael Pittman Jr. | WR | 29 | 0.93 | 94 | 60 | +34 | 2 | 1 |
| Wan'Dale Robinson | WR | 26 | 0.92 | 117 | 84 | +33 | 3 | 1 |
| Alec Pierce | WR | 26 | 0.92 | 86 | 56 | +30 | 3 | 3 |
| Jakobi Meyers | WR | 30 | 0.90 | 124 | 95 | +29 | 4 | 1 |
| Dalton Schultz | TE | 30 | 0.93 | 95 | 67 | +28 | 2 | 2 |
| Tony Pollard | RB | 29 | 0.94 | 83 | 57 | +26 | 1 | 0 |
| Bo Nix | QB | 26 | 0.95 | 65 | 42 | +23 | 0 | 0 |
Hurt most — the projection assumes a season he has not been having
| Player | Pos | Age | Avail | Board rank | Adjusted rank | Move | Games missed, 3yr | Weeks listed Out |
|---|---|---|---|---|---|---|---|---|
| Cam Skattebo | RB | 25 | 0.65 | 26 | 108 | -82 | 9 | 0 |
| Rashee Rice | WR | 26 | 0.62 | 25 | 103 | -78 | 23 | 2 |
| Malik Nabers | WR | 23 | 0.66 | 47 | 118 | -71 | 15 | 2 |
| Joe Burrow | QB | 30 | 0.72 | 57 | 124 | -67 | 16 | 3 |
| Jayden Daniels | QB | 26 | 0.75 | 54 | 116 | -62 | 10 | 10 |
| Christian Watson | WR | 27 | 0.70 | 68 | 120 | -52 | 18 | 4 |
| Omarion Hampton | RB | 24 | 0.68 | 14 | 66 | -52 | 8 | 2 |
| Tyler Shough | QB | 27 | 0.74 | 80 | 125 | -45 | 6 | 0 |
| Brock Purdy | QB | 27 | 0.80 | 90 | 117 | -27 | 11 | 7 |
| Tee Higgins | WR | 28 | 0.79 | 69 | 96 | -27 | 12 | 5 |
Position mean availability used for the adjustment
These are the divisors in Adj VOR — the mean Avail across every row on the board at that position, rookies included. They are deliberately not the same as the position base rates build_context.py prints, which are computed only over players with NFL history. Rookies carry the base rate as their Avail, so folding them back into this mean pulls it slightly toward the base rate and makes the adjustment marginally gentler. Both numbers are correct; they answer different questions.
| Pos | Mean Avail (adjustment divisor) |
|---|---|
| QB | 0.841 |
| RB | 0.813 |
| WR | 0.850 |
| TE | 0.840 |
Open questions
- The survivor selection above is the real limit on this note. Measuring the wider claim needs a pool defined before the injuries, not after.
Availtreats a missed game as a missed game, whether it cost a start in week 3 or in week 16. A playoff-weeks-weighted version would matter more to a league whose money is decided in weeks 15–17.- Nothing here predicts this season's injuries, and three seasons is a thin sample for a rate. It prices what a player has done, not what he will do.