Injury-prone isn't a myth. It's a three-game tax, and only on repeat offenders.
What this means for your roster. The injured discount is real money, but only on the first one. Buy Xavier Worthy at WR40, where we have 14.5 expected games and the room has him cheaper than we do. Hold Javonte Williams at RB17 and Christian Watson at WR32, two guys the injury label has followed around for no charge. Then stop paying the same price twice: fade Ricky Pearsall at WR64 (9.2 expected games), Cam Skattebo at RB33 (8.0), and J.K. Dobbins at RB61 (10.2). Malik Nabers at WR54 is where we sit furthest from the room, and that is our recovery floor talking about his calendar, not a doctor talking about his knee.
What the consensus says
August is buy-low season and the genre has one setting. PlayerProfiler frames the whole edge as ignoring the fear baked into the market: while others hesitate, you buy the discounted veteran coming off an injury. RotoBaller runs seven of them. FantasyPros and SI run their own. DraftSharks makes the cleanest version of the argument, that receivers return to prior production after a season-ending injury more reliably than backs do, and names Christian Watson while conceding he has missed 20 games in four seasons.
The through-line: an injury is a discount, not a disclosure. Nobody in that stack separates the guy who got hurt once from the guy who keeps getting hurt.
The claim, in plain English
Take every season where a fantasy-relevant player was hurt enough to miss real time but still played, between 30 and 75 percent of his team's games. Then look at what was already on his card. If he had another shortened season in the previous three years, he is a repeat. If his last three years were clean and he had a full season somewhere behind him as proof he can do it, he is a first-timer.
Compare the two groups the next season, holding fixed how good he was per game in the year he got hurt. That control is the whole thing, because otherwise you are just discovering that better players score more. The instrument is scripts/h82_injury_recurrence_availability_floor.py.
How I beat on it
547 qualifying player-seasons, quarterbacks through tight ends, 2015 through 2024, next season scored in half PPR with a tight end premium. I score the following year as points per team game, dividing by the full 17 rather than by games he suited up for, because your lineup eats the empty weeks either way. Rookies hurt in their first year are a separate case and sit this one out, 109 of them. Public stat file only. Framework in the research spec. Nothing from 2026 touches it.
What the data actually said
First swing, no controls, and it is not close. First-time missers came back at 74.8 percent availability. Repeats came back at 57.5. Call it 2.9 games out of seventeen, gone before the season starts.
Because aggregates lie, I cut it by how productive each guy was in the year he got hurt. Five bands, from elite down to flex-adjacent, and the gap is negative in all five: 1.80, 3.20, 1.77, 2.19 and 2.01 points per team game against the repeats. There is no band where the pattern washes out, which is what you would see if this were really a talent gap wearing an injury costume.
| Y0 points per game | Repeat, next yr | First-time, next yr | Gap |
|---|---|---|---|
| 16 and up | 10.07 | 11.87 | -1.80 |
| 13 to 16 | 7.75 | 10.95 | -3.20 |
| 10 to 13 | 5.31 | 7.08 | -1.77 |
| 7 to 10 | 3.78 | 5.97 | -2.19 |
| 5 to 7 | 2.95 | 4.96 | -2.01 |
Overall it is 3.16 points per team game. And that is the polite number: 20.8 percent of the repeats never recorded a snap the following season at all, against 10.2 percent of the first-timers. Count those seasons as the zeroes they were and the gap widens to 3.30.
The position split is where it gets useful. Running backs, the position the injury-prone label gets shouted at loudest, show almost no availability gap between repeats and first-timers, 63.9 percent against 67.6, about six-tenths of a game. Their damage lands on the rate instead, 6.13 against 8.38 a game. Receivers and tight ends are the opposite, losing three and four games of calendar respectively. So the label is being applied hardest at exactly the position where it means the least.
What the engine already figured out
This one is a rare handshake, and I checked it in the direction that could have embarrassed us. SignalTuned's availability floor already splits these two groups, and it already refuses to floor the repeats. It carries 0.560 for a repeat and 0.726 for a first-time miss with healthy proof behind it, a spread of 2.82 games. The cohort says 0.575 and 0.748, a spread of 2.94. The calibration is off by a tenth of a game on a number that decides how much of a star's price you are willing to pay. Nothing to change.
I also audited whether the floor is actually firing rather than merely existing, which is not the same question. Across 85 players currently flagged as a single acute event, every one sits at or above the floor, with four backup quarterbacks capped by the depth chart on purpose. Zero misses. Last month's post found the injury tax lands on the calendar and not on production. This is the other half: whose calendar stays broken.
The 2026 names
Xavier Worthy, WR40, age 23.3, 14.5 expected games. Buy. Above-average durability, the market has him cheaper than we do, and there is no pattern on the card.
Javonte Williams, RB17, age 26.4, 15.5 expected games. Hold. A single acute event years behind him and full availability since. This is what a healed one looks like.
Christian Watson, WR32, age 27.3, 11.0 expected games. Buy. The one the buy-low columns keep flagging as too fragile. He is a repeat and we still like him, because the discount finally got bigger than the tax.
Rome Odunze, WR23, age 24.2, 13.2 expected games. Hold. One foot, no pattern, and the room has him about where we do. Not every consensus call is wrong.
Ricky Pearsall, WR64, age 26.0, 9.2 expected games. Fade. The market is buying a first-time bounce on a card that no longer reads that way.
Cam Skattebo, RB33, age 24.6, 8.0 expected games. Fade. A first-year miss with no healthy season behind it, the branch our floor deliberately leaves alone.
J.K. Dobbins, RB61, age 27.7, 10.2 expected games. Fade. Chronic pattern, and running backs pay theirs in efficiency rather than in games.
Jayden Daniels, QB14, age 25.7, 12.3 expected games. Fade at the room's price. He got the floor, we lifted him for it, and he is still short of what he costs.
Rashee Rice, WR59, age 26.4, 11.1 expected games. Fade. Floored, not forgiven.
Malik Nabers, WR54, age 23.1, 11.1 expected games. Fade at market. This is our largest disagreement with the room and it deserves the plain version: we are not calling his knee, we are refusing to pay a full season's price for eleven games. If he plays sixteen, we were wrong and you got him cheap from someone else.
Joe Burrow, QB8, age 29.7, 10.8 expected games. Hold. A repeat by our own definition, priced accordingly, and still a top-ten superflex asset because the rate is that good.
Tucker Kraft, TE7, age 25.8, 10.1 expected games. Hold. Elite when upright. We are not paying the tight end premium twice for the weeks he is not.
What to do about it
Ask one question before you send the offer: is this his first one? A first-time miss with a healthy season behind him is a discount on a player. A second one is a disclosure about a calendar, and it costs you three points a week you will never see on a box score because it shows up as an empty roster spot.
No engine change ships from this. The calibration was already right, verified against outcomes rather than against itself, and a change nobody can defend is worse than no change. What does go on the list is the position split: our floor is one flat pair of numbers for all four positions, and the cohort says backs and receivers are not paying the same tax in the same currency. Quarterbacks show the widest gap of all, and I do not trust it yet, because a quarterback who misses time often loses the job rather than the leg, and that is a depth chart problem wearing an injury's clothes.
Receipts
Cohort output data/h_injury-recurrence-availability-floor_1786554363.json. Floor coverage audit docs/receipts/g50_acute_floor_coverage_2026-08-12.json. Board ranks from the 12-team superflex half-PPR tight-end-premium export, engine v1.6.144.