You're chasing the wrong WR metric. Air yards share does not predict Y+1 production.
At equal target share, don't pay a premium for the WR who earns disproportionate air yards. Two seasons, 139 WR pairs: no measurable Y+1 advantage for the vertical specialist over the route hog. When choosing between two WR1s at equal dynasty price and similar target share, the air yards allocation is not the tiebreaker it looks like.
What the consensus says
Dynasty discourse prices air yards leaders as value. Dynasty Nerds ranks air yards allocation among the top WR forward metrics, alongside target share and route participation. PFF's fantasy coverage regularly flags disproportionate air yards share as a breakout indicator. ESPN's air yards tracker built careers for analysts who framed disproportionate air yards as a WR's "real" opportunity signal -- a WR earning a larger slice of team air yards than her slice of targets runs deeper, harder-to-cover routes. The QB trusts her in the vertical passing game. The Athletic's dynasty coverage routinely surfaces disproportionate air yards as evidence of ceiling potential.
The conventional wisdom is seductive. Air yards do measure something real about WR usage -- a WR getting disproportionate air yards runs a deeper route tree; her targets are worth more raw yards when they connect. But real does not mean predictive of next-year output.
The claim, in plain English
The predicate: a WR is "disproportionate air yards" if her share of team air yards exceeds her share of team targets by at least 20 percent.
def predicate(row: pd.Series) -> bool:
ts = float(row.get("target_share", 0) or 0)
ays = float(row.get("air_yards_share", 0) or 0)
if ts <= 0.05:
return False
return (ays / ts) >= 1.20
Tyreek Hill in 2022 earned 30 percent of Dolphins targets and 39 percent of team air yards. Ratio: 1.30 -- below the threshold. His air yards are roughly proportional to his target load. Mike Evans in 2023 earned 25 percent of Buccaneers targets and 40 percent of team air yards. Ratio: 1.61 -- above the threshold. Evans runs deeper. By the discourse logic, Evans earns the forward premium. The question: does it show up in next-year production?
The control band is target share. If the air yards signal adds something, it must survive within-target-share bands -- at equal Y0 target share, the disproportionate-air-yards WR must produce more Y+1 ppg. Without that control, we're measuring the correlation between air yards and target share, which everybody already prices in.
How I beat on it
scripts/h70_wr_air_yards_share_premium.py aggregated 2022-2024 weekly data to season level, filtered WRs to at least 6 games and 40 targets in Y0, and built Year-to-Year+1 pairs for 2022 and 2023. 139 pairs total. Scoring: Dynasdeez 10-team superflex Half-PPR. Framework: two-gate spec from Session 97.
What the data actually said
First swing without the target-share control: WRs with a high air-yards/target-share ratio (above the sample median of 1.33) produced 8.88 ppg in Y+1. WRs below the median produced 10.13 ppg. A -1.25 ppg gap going the wrong direction for the hypothesis.
That aggregate is partly confounded by low-volume WRs. A backup receiver who earns 0.10 target share but 0.24 air yards share (ratio 2.4) is getting deep shots on a bad team -- a high ratio, low production floor. So we control for target share:
| Target-share band | High ays/ts ratio Y+1 ppg | Low ays/ts ratio Y+1 ppg | Delta |
|---|---|---|---|
| ts >= 0.25 (alpha WR1s) | 12.57 ppg (n=19) | 12.96 ppg (n=18) | -0.39 |
| 0.20-0.25 | 9.10 (n=19) | 10.56 (n=19) | -1.46 |
| 0.16-0.20 | 8.83 (n=14) | 7.26 (n=14) | +1.57 |
| 0.12-0.16 | 6.69 (n=13) | 6.53 (n=13) | +0.16 |
Gate 1 requires a consistent monotonic direction across bands -- and this data does not have one. The signal is negative in two bands, positive in one, flat in one. Within the band that matters most for dynasty decisions (ts >= 0.25, where WR1 debates live), the delta is -0.39 ppg -- noise.
Mike Evans and A.J. Brown, both above the median ays/ts ratio in 2023, produced 14.5 and 14.1 ppg in 2024. Tyreek Hill and CeeDee Lamb, both below the median in 2022, produced 19.8 and 19.7 ppg in 2023. The examples land the narrative. But Ja'Marr Chase -- also above the median in 2023 -- jumped from 13.3 to 20.0 ppg. Individual seasons are noise. Across 139 pairs, the signal is absent.
Gate 1 fails.
What the engine already figured out
The engine does not encode air_yards_share / target_share as a ranking multiplier, and this test confirms that is correct. Air yards share appears as a descriptive chip on every WR page. The PWOPR module flags WRs with inflated air yards as REGRESSION RISK -- a downward signal, not a premium. A prior harness verdict found that rewarding high-air-yards WRs via a projection bonus breaks the top-5 WR named cases (every consensus elite WR shows a persistent positive WOPR residual that snap-count proxies can't distinguish from genuine scheme advantage). The lift stays off. Descriptive-only is the right disposition.
What to do about it
The WR acquisition signal that predicts Y+1 production is target share. Not air yards share. Not what percentage of team air yards that target share converts into. Not aDOT.
At equal target share, Hill and Lamb's proportional route trees produced at least as much as Evans and AJ Brown's vertical-heavy trees. The mechanism is plausible -- short-area possession routes stay efficient longer than deep separations that erode with age and defensive attention -- but with 37 WR1 pairs across two years, this is a direction, not a verdict.
What is clear: the air yards premium, the forward signal dynasty discourse builds buy cases around, does not materialize in next-year production when you actually control for target share. Don't pay it.