injury return pathway
how the model prices players returning from IR
the standard projection model has a blind spot: players who were on IR for most of 2025 have no recent production signal. a model that reads recent weekly stats and finds nothing will either produce a garbage projection from stale pre-injury data or skip the player entirely.
both outcomes are wrong. we built the injury-return pathway to fix them. this page explains what it does and how to read its output. the calibration behind it stays proprietary.
the problem
a player who missed most of 2025 looks like noise to the base model: a tiny, high-variance sample that produces a low-confidence projection and ranks the player far below their real expected value. that is a systematic error, and it's the error the pathway corrects.
what the pathway does
for each affected player, the pathway prices the return rather than ignoring it. it accounts for the type of injury and how much recovery uncertainty comes with it, discounts production toward what players in similar situations have actually done on return, and blends that against whatever recent signal exists. the more uncertain the return, the more the projection leans on a conservative anchor instead of an optimistic guess.
the important part for you is the output: a projected FP/game and a confidence score that honestly reflect both what we know and what we don't. the exact discounts and tiers are calibrated on historical return-game data and are not published.
confidence is the signal
the model expresses return uncertainty primarily through the confidence score, not through an inflated projection. a low confidence score on a returning player is the model saying "this is a wide range, not a point estimate." that is a feature, not a missing number.
this is conservative by design. overconfidence on injury returns has a known failure mode: the first game back goes badly (re-aggravation, load management, rust), the projection was way off, and you've been rostering a liability. the pathway would rather give you an honest range than a confident miss.
the 58 players
the 2026 version of the pathway covers 58 players returning from IR or significant injury. every one is flagged with data_source = "injury_return" in the API response, so you can filter the rankings to see the full cohort or pull any one of them by ID.
what it does not do
- it does not predict whether a player will be re-injured. nobody does that reliably.
- it does not account for load management decisions (coaches easing returning players back). these depress real return-game production but are unknowable in advance.
- it does not pretend its samples are bottomless. some injury types have very little history, and the confidence score reflects that.
the pathway is honest about its uncertainty. that's the whole point: a low confidence score means "this is a range with high variance," not "skip this player."
frequently asked questions
can I filter the API to see only injury-return players?
yes. query /v1/rankings/preseason and filter on data_source == "injury_return", or request a specific player by ID.
why is my favorite player's confidence so low?
low confidence on a returning player reflects the uncertainty of the return, not a judgment about the player's ability. it's the model saying "this is a range," and the range is wide because returns are genuinely hard to call.
how is this different from just projecting the pre-injury number?
projecting the raw pre-injury production without adjustment systematically overstates return-game output. even routine returns tend to come back below prior form for a stretch. the pathway prices that in.