'Injury-prone' is a quarterback word. You keep spending it on running backs.
What this means for your roster. Fade Joe Burrow. We have him QB20 and the room is taking him 14th overall, and our expectation for him is 10.8 games. Fade Kyler Murray at QB31, 8.8 expected games. Then stop paying the injury tax twice at running back: hold De'Von Achane at RB4, who lands on every high-risk list and grades out on our board as elite durable at 15.9 expected games, and hold Breece Hall at RB8 with an ACL already on his card and 15.5 expected games. Christian McCaffrey at RB42 is still a fade against an ADP of 32, but sell him for the birthday, not the injury history. And when you do fade a repeat-miss back, know what you are buying: he is going to play. He is just going to be worse.
What the consensus says
Pull up any fade column published this month. SI runs "4 running backs that are high injury-risks." RotoBaller calls them RB landmines. FantasyPros has five RB busts, RotoWire has four more, CBS explains dynasty turnover at the top as "injuries and age," and DraftSharks runs a whole injury predictor with a dedicated RB page. McCaffrey is the headliner, ten career injuries. Achane is on the list for seven in four seasons.
Six sweeps, six lists, and not one of them applied the label to a quarterback.
The claim, in plain English
The claim is that a player who has already missed time is more likely to miss it again, and that this is a running back problem. So here is the test. Take every fantasy-relevant player whose season was cut to somewhere between 30 and 75 percent of the schedule. Split them into two piles: guys who had already lost a chunk of a season in the previous three years, and guys for whom this was the first one and who have a full healthy season on the record. Then look at what each pile does the next year. Same predicate my previous run used, same cohort, imported rather than rewritten so the two are comparable line for line. The script is here.
The only new thing is that this time I cut it by position and put a confidence interval on every branch, resampling players rather than seasons, so one guy with eight injury years counts as one guy.
How I beat on it
Y0 to Y+1 pairs, 2015 through 2024, half-PPR. 547 injury-shortened seasons, 451 with a next season to grade, 379 distinct players. Ten thousand cluster-bootstrap draws per branch. The framework is here.
What the data actually said
First swing, everybody pooled: repeat missers come back at 57.5 percent availability, first-timers at 74.8. A gap of 17 points, which is 2.9 games of a 17-game season. That reproduced my earlier run to the third decimal.
Because pooled numbers lie. Cut it four ways:
| Position | Repeat misser | First-timer | Gap | 95% range |
|---|---|---|---|---|
| QB | 43.3% | 69.6% | -26.3 | -38.3 to -14.5 |
| TE | 63.9% | 87.9% | -24.0 | -36.4 to -12.5 |
| WR | 59.8% | 78.1% | -18.3 | -26.5 to -10.4 |
| RB | 63.9% | 67.6% | -3.6 | -15.1 to +8.1 |
Three positions sit on top of each other, right where the pooled number is. Running back is not near them. Its range runs from minus fifteen to plus eight, which is a polite way of saying I cannot tell you the sign. And this is not a sample problem dressed up as a finding: running back has the second-largest player count of the four. To resolve that gap against zero I would need 982 players and I have 97. The effect is not hiding. It is small.
The quarterback number is the one I tried hardest to kill. A quarterback who stops playing might have lost the job rather than the leg, and that is a different thing entirely. So I threw out every season with no injury actually on the report and ran it again on injuries only. The gap came back at minus 26.2. It moved by a tenth of a point.
Now the part that pays your rent. Repeat-miss running backs lose 2.25 points of per-game scoring, and that range does not cross zero. They played 8.8 games at 8.2 points. First-timers played 9.5 games at 9.8. The back does not sit out. He suits up and he is worse. Everybody else pays on both counts.
What the engine already figured out
Our recovery floor carries one recurrence number for all four positions, and my last run said that number was almost exactly right: encoded 56.0 percent against 57.5 observed. This run says the average was right and the average was hiding two things.
Translated into points, six of the eight position branches sit inside tolerance, including both running back branches and both receiver branches. Two do not. We are too generous to repeat-miss quarterbacks by 1.27 points per team game, because a quarterback scores 10 to 16 a game and the same missing week costs triple what it costs a 6-point back. And we are too harsh on tight ends coming back from a single miss by 1.22. Fixing either one takes a move of roughly 20 percent, four times what a single research cycle is allowed to propose, so it goes to the calibration loop with a real harness rather than getting nudged here. Our ranks do not move today, and that is the one-line reason.
Where we already get it right is on players with a long record, because the engine prices your actual history before it reaches for any group average. That is why Achane grades elite durable while the lists call him fragile.
What to do about it
| 2026 name | Our board | Call |
|---|---|---|
| Joe Burrow | QB20 | Fade, 10.8 expected games and the room is drafting him 14th overall |
| Kyler Murray | QB31 | Fade, 8.8 expected games, three shortened seasons in four |
| Deshaun Watson | QB95 | Fade, 2.5 expected games, there is no price we like |
| Anthony Richardson | QB40 | Hold at his price, we already have him at 2.5 games as a backup |
| Trevor Lawrence | QB8 | Hold, 15.6 expected games, one bad year is not a pattern |
| Justin Herbert | QB12 | Hold, 16.1 expected games, the most boring good thing you can own |
| De'Von Achane | RB4 | Hold, elite durable at 15.9 games, the lists are wrong about him |
| Breece Hall | RB8 | Hold, 15.5 expected games with an ACL on the card |
| Christian McCaffrey | RB42 | Fade against ADP 32, on the birthday, not the knee |
| Cam Skattebo | RB33 | Fade, 8.0 expected games, we are well under the room |
| J.K. Dobbins | RB61 | Fade, and expect him to play badly rather than not play |
The trade to go make: your league-mate has an injury-prone note next to a running back and no note at all next to his quarterback. He has that exactly backwards. Buy the back at the discount, and never take his quarterback at par.
One honest caveat: the injury report is the only tool I have for telling a benching from a real injury, and it covers only 44 percent of quarterback miss-seasons, the thinnest coverage of the four positions. The quarterback gap survived that filter unchanged, but it survived on 47 seasons, so treat the size of it as directionally solid and the last decimal as soft.