The intuition is that two receivers on the same team split a fixed pie, so one’s big day is the other’s quiet one. Measured on raw outcomes over full games, it runs the other way.
What moves together
| Pair | Correlation |
|---|---|
| Two pass catchers, same team | +0.328 |
| Quarterback and his own receiver | +0.253 |
| Opposing passing games (a shootout) | +0.172 |
| Two backs, same team | +0.122 |
| Players in different games | 0 |
Across a whole game the shared environment dominates. A team that throws 45 times because it fell behind, or because the game turned into a shootout, feeds every receiver it has. The pie isn’t fixed. Its size is the biggest thing that varies.
The measurement that nearly hid it
An earlier cut measured correlation between “did he beat his own season average” flags and found +0.002, essentially nothing. That proxy subtracts each player’s own baseline, and the shared game environment is exactly what the subtraction removes. Raw outcomes are the right unit, and the gap between the two numbers is the finding.
What it means for a projection
Two receivers on one team are more likely to have a big day, or a quiet one, together than projecting each on his own suggests. Treating them as independent understates how often their outcomes land together. The practical point runs the other way from the intuition: pairing teammates in a fantasy lineup concentrates it on one game script, it doesn’t spread it.