How much does the opponent actually matter for a player prop projection?
Much less than intuition suggests — scaling a player's own average by the opponent's tendency changed the projection's average error by well under 2% in every market tested, and only clearly helped in 3 of 4.
See a real opponent matchup in Player Props
The evidence
Projection error (mean absolute error, in the stat's own units): own-average-only versus opponent-ratio-adjusted
| Market | No opponent term | Opponent-adjusted | Games |
|---|---|---|---|
| WR receiving yards | 25.34 | 25.32 | 9045 |
| RB rushing yards | 24.19 | 24.58 | 5440 |
| QB passing yards | 65.4 | 65.18 | 2686 |
| TE receiving yards | 17.45 | 17.41 | 4623 |
How this was measured
For every player-game with at least 8 prior games of history (no lookahead — only games before the one being predicted), compared two projections against the actual value: the player's own cumulative prior average, and that average scaled by the opponent's allowed rate versus the league's prior rate at the same point in time (this app's own opponent-ratio method). Error is mean absolute error in each stat's own units (yards).
Sample size: 21,794 player-games (across 4 markets). Last verified 2026-09-27.
What this doesn't tell you
This tests one specific opponent-adjustment method (a ratio scaling) against one specific baseline (the player's own average) — it doesn't rule out that a different, more sophisticated opponent model could do better. It also measures average prediction error, not betting profitability, and says nothing about any single matchup.
Related questions
This is a place to look, not a bet to make — every figure here describes the past, not a specific upcoming game.