The predictor ladder · GWTTKB Research

Volume persists. Efficiency reverts. The full sweep, season by season

Every stat field in the database tested against next-period scoring, five seasons, with the replication flag on each. Target share carries. Yards per target runs negative.

GWTTKB Research · updated 2026-08-15
Free to read. No account, no paywall.

The question

What actually predicts fantasy football production — volume or efficiency?

What the field generally says

Fantasy analysis leans heavily on efficiency — yards per target, yards per carry, yards per route run — as evidence of who is good and therefore who will score.

What we measured

Opportunity measures replicate across every season tested. Efficiency measures mostly do not, and several are negative forward predictors: a player who was hyper-efficient on low volume tends to come back down, so a gaudy per-touch number is a warning rather than a selling point.

Wide receivers and tight ends

Target share+0.247WOPR+0.218Air-yards share+0.142Cushion+0.139Separation+0.128Receiving EPA-0.127ydsPerTgt-0.114Week-to-week volatility-0.075fantasyForm+0.050teamImplTrend-0.050Pace trend-0.046racr-0.0460
Every factor against next-period scoring. Green replicates positive, red replicates negative, grey failed to hold its sign across all seasons tested. n = 4,488 player-stretches.
Target shareWOPRAir-yards shareCushionSeparationReceiving EPAydsPerTgtWeek-to-week volatility0
The replication bar, made visible. One dot per season tested. A factor only counts when every dot sits on the same side of zero — scattered dots are the failures.

Running backs

Week-to-week volatility-0.163Receiving EPA-0.143Efficiency+0.135Time to line of scrimmage+0.128Target share+0.119snapTrend+0.103Rushing EPA-0.082fantasyForm+0.078Catch rate+0.063stackedBox+0.062carriesTrend+0.057Pace trend-0.0540
Every factor against next-period scoring. Green replicates positive, red replicates negative, grey failed to hold its sign across all seasons tested. n = 2,022 player-stretches.

Quarterbacks

Team implied total+0.146CPOE-0.091Pace trend-0.085Aggressiveness-0.077rushFloor+0.072timeToThrow-0.063schedSoftAhead+0.061passEpa+0.051snapTrend+0.045Week-to-week volatility+0.038teamPassEpa+0.037intendedAir-0.0330
Every factor against next-period scoring. Green replicates positive, red replicates negative, grey failed to hold its sign across all seasons tested. n = 1,151 player-stretches.

Why it comes out that way

Volume is assigned by a coaching staff and is stable week to week because the reasons for it — role, scheme fit, trust — are stable. Efficiency is produced partly by the player and partly by coverage, blocking and where on the field the opportunity started, and those inputs reshuffle every week.

How it was tested

Each factor is correlated with next-period fantasy scoring separately in each season. A factor is marked as replicating only when it holds the same sign in every season tested — one good season is a coincidence. The charts below are read at build time from the deployed sweep, so they are the same numbers the site’s projections use.

Figures on this page are read at build time from the same data file the site’s projections use.

Two figures, and why they differ

Two sweeps are published and they are not the same measurement. The charts here group wide receivers with tight ends across 4,488 player-stretches and put target share at +0.247. The methodology report runs wide receivers alone against the next four weeks specifically, across 3,222 stretches, and puts target share at +0.217. Different population, different forecast window, same conclusion — opportunity replicates and efficiency does not. If you are citing a figure, cite the one whose population matches your question.

What this does not say

These are population-level correlations across qualifying player-stretches. They describe what to trust when ranking a field of players, not a guarantee about any individual. A correlation of +0.25 is a strong fantasy signal and still leaves most of the variance unexplained.

Ask the Coach about a player →
Does rushing yards over expected predict fantasy football production?Do fantasy football touchdowns regress? The full methodology report — 272 fields tested, 41 replicated
GWTTKB publishes research and education. Nothing here is betting advice.
All findings · 172 player studies · Methodology · Research index