New to these reports? Start here
- Dotted-underlined words have a plain-English definition — hover or tap them. Every term is also on the glossary page.
- "Null" means we found nothing, not that something broke. Most reports here are negative results, on purpose — knowing an idea doesn't work is the point.
- Two questions get asked separately. First, is the effect real? Second, is it already priced into the betting odds? An effect can be completely real and still useless to bet on.
- A "calibration" row is a self-check. It runs the same method on something already known to be true. If that fails, the whole report is unreliable — so it's reported alongside the findings.
- If a confidence interval includes zero, the real effect might be nothing at all, so no claim gets made.
Fantasy points after a big loss
Question (2026-10-29): do players score more PPRpoints per receptionA fantasy scoring format that awards a point for every catch, which raises the value of high-volume receivers. in the game after their team loses by 10+, adjusting for who beat them?
Answer: no. Players score slightly less: about 0.3 PPR below their own average (90% CI −0.46 to −0.11) after adjusting for the loss opponent, the next opponent's defence against the position, and home/away. That's ~2-3% of a typical fantasy starter's game. There is no bounce-back.
Method (bounceback.py)
- 2012-2025 regular season, QB/RB/WR/TE who played both the loss and the next game for the same team; leave-out season average ≥ 5 PPR over ≥ 4 other games. 35,763 player-games, 7,999 after a 10+ point loss.
- Outcome: PPR minus the player's season average in his other games (neither the loss nor the next game counts toward it).
- Loss opponent: its average point margin in its other games that season (also tried: the closing spread of the loss). Next opponent: PPR it allowed to the position, leave-out, minus league average. SEs clustered by team-game.
Results (effect of a 10+ point loss, PPR)
| Model | Effect | 90% CI |
|---|---|---|
| Unadjusted | −0.30 | −0.47 to −0.13 |
| + next opponent, home | −0.32 | −0.49 to −0.15 |
| + loss opponent strength | −0.29 | −0.46 to −0.11 |
| + loss opponent via closing spread | −0.36 | −0.54 to −0.18 |
| + next game's implied team total | −0.17 | −0.35 to +0.01 |
- Who beat them barely matters: big loss to a weak / middling / strong team −0.37 / −0.27 / −0.26.
- By position: QB −0.49, WR −0.45, TE −0.37, RB +0.14 (n.s.).
- By margin of previous game (mean residual): lost 20+ −0.25, lost 10-19 −0.32, lost 1-9 +0.12, won 1-9 −0.15, won 10+ +0.08. Not monotonic; the bands are within noise of each other.
- Halves: 2012-2018 −0.50; 2019-2025 −0.06 (n.s.).
Reading it
- The small dip is mostly the team being worse than its season average says: adding the market's implied total for the next game cuts it to −0.17 and makes it insignificant. So the betting market already expects it; there's no prop edge here.
- It isn't stable: only the first half of the sample shows it.
- Caveat: the sample requires playing both games, so players hurt in the blowout drop out; backups who replaced them are in only if they clear the 5-PPR baseline.
Data: nflverse weekly player stats and schedules (data/, gitignored-size parquet).
On this page
Terms in this report
Source
backtests/nfl_bounceback/FINDINGS.md
updated 2026-10-29 23:51