NFL · Empirical Bayes
True-Talent Leaderboards
A raw single-season rate is a noisy guess at a player’s real ability, and the noise is worst for the smallest samples — so a naïve leaderboard is topped by whoever got lucky in the fewest tries. This one shrinks every rate toward its position-group prior by how much the sample can be trusted, and shows a 90% credible interval instead of a point. Toggle raw vs. shrunk to watch the flukes fall back to the pack.
Dropback success rate · 2024
Share of dropbacks (incl. sacks & scrambles) with positive EPA.
- Beats raw by
- 15.6%
- lower out-of-sample error
- RMSE raw → shrunk
- 5.5% → 4.6%
- odd vs. even weeks
- Split-half reliability
- 0.47
- how repeatable the raw stat is
- Stabilizes at
- 121
- dropbacks to trust the number
Each row is a player: the solid dot is the shrunk estimate, the hollow dot the raw rate, joined by the red pull of regression; the grey bar is the 90% credible interval. A hollow shrunk dot means the sample is below the stabilization line — the number is mostly the position prior. Switch to Raw rank and watch the small-sample names climb.
Dropback success rate · 2024 · full board
| # | Player | Team | Pos | dropbacks | Raw | Shrunk | 90% interval |
|---|---|---|---|---|---|---|---|
| 1 | Jared Goff | DET | QB | 568 | 54.0% | 52.3% | 49.2% – 55.5% |
| 2 | Tua Tagovailoa | MIA | QB | 421 | 53.0% | 51.0% | 47.5% – 54.6% |
| 3 | Lamar Jackson | BAL | QB | 494 | 52.0% | 50.5% | 47.2% – 53.8% |
| 4 | Joe Burrow | CIN | QB | 701 | 51.2% | 50.2% | 47.3% – 53.1% |
| 5 | Baker Mayfield | TB | QB | 612 | 51.3% | 50.1% | 47.1% – 53.2% |
| 6 | Kyler Murray | ARI | QB | 570 | 49.8% | 48.9% | 45.7% – 52.0% |
| 7 | Brock Purdy | SF | QB | 482 | 49.6% | 48.5% | 45.2% – 51.9% |
| 8 | Patrick Mahomes | KC | QB | 617 | 49.1% | 48.3% | 45.3% – 51.3% |
| 9 | Josh Allen | BUF | QB | 497 | 48.9% | 48.0% | 44.7% – 51.3% |
| 10 | Sam Darnold | MIN | QB | 593 | 48.6% | 47.8% | 44.8% – 50.9% |
| 11 | Matthew Stafford | LA | QB | 546 | 48.4% | 47.6% | 44.4% – 50.8% |
| 12 | Kirk Cousins | ATL | QB | 478 | 48.1% | 47.3% | 44.0% – 50.7% |
| 13 | Marcus Mariota◦ provisional | WAS | QB | 47 | 53.2% | 46.8% | 40.5% – 53.1% |
| 14 | Tanner McKee◦ provisional | PHI | QB | 47 | 53.2% | 46.8% | 40.5% – 53.1% |
| 15 | Joshua Dobbs◦ provisional | SF | QB | 48 | 52.1% | 46.5% | 40.2% – 52.8% |
| 16 | Jordan Love | GB | QB | 442 | 47.1% | 46.5% | 43.0% – 49.9% |
| 17 | Michael Penix Jr.◦ provisional | ATL | QB | 109 | 48.6% | 46.3% | 40.9% – 51.8% |
| 18 | Tyrod Taylor◦ provisional | NYJ | QB | 25 | 56.0% | 46.3% | 39.5% – 53.1% |
| 19 | Mason Rudolph | TEN | QB | 237 | 47.3% | 46.3% | 41.9% – 50.6% |
| 20 | Geno Smith | SEA | QB | 629 | 46.4% | 46.1% | 43.1% – 49.1% |
| 21 | Joe Milton III◦ provisional | NE | QB | 29 | 51.7% | 45.7% | 39.1% – 52.4% |
| 22 | Jayden Daniels | WAS | QB | 528 | 45.8% | 45.5% | 42.3% – 48.8% |
| 23 | Jimmy Garoppolo◦ provisional | LA | QB | 45 | 48.9% | 45.5% | 39.2% – 51.9% |
| 24 | Joe Flacco | IND | QB | 267 | 46.1% | 45.5% | 41.4% – 49.7% |
| 25 | Drake Maye | NE | QB | 374 | 45.7% | 45.4% | 41.7% – 49.1% |
| 26 | Jalen Hurts | PHI | QB | 399 | 45.6% | 45.3% | 41.7% – 48.9% |
| 27 | Justin Herbert | LAC | QB | 545 | 44.8% | 44.7% | 41.5% – 47.9% |
| 28 | Gardner Minshew | LV | QB | 333 | 44.7% | 44.6% | 40.8% – 48.5% |
| 29 | Derek Carr | NO | QB | 290 | 44.1% | 44.2% | 40.2% – 48.2% |
| 30 | Aidan O'Connell | LV | QB | 254 | 44.1% | 44.1% | 40.0% – 48.4% |
| 31 | Aaron Rodgers | NYJ | QB | 627 | 43.4% | 43.5% | 40.6% – 46.5% |
| 32 | Trevor Lawrence | JAX | QB | 305 | 43.0% | 43.3% | 39.4% – 47.3% |
| 33 | Mitchell Trubisky◦ provisional | BUF | QB | 26 | 38.5% | 43.2% | 36.6% – 50.0% |
| 34 | Bo Nix | DEN | QB | 591 | 42.6% | 42.9% | 39.9% – 46.0% |
| 35 | Daniel Jones | NYG | QB | 371 | 42.3% | 42.8% | 39.1% – 46.5% |
| 36 | Malik Willis◦ provisional | GB | QB | 63 | 39.7% | 42.7% | 36.7% – 48.7% |
| 37 | Drew Lock | NYG | QB | 193 | 41.4% | 42.5% | 38.0% – 47.1% |
| 38 | Justin Fields | PIT | QB | 177 | 41.2% | 42.5% | 37.8% – 47.2% |
| 39 | Tommy DeVito◦ provisional | NYG | QB | 50 | 38.0% | 42.4% | 36.3% – 48.7% |
| 40 | Andy Dalton | CAR | QB | 168 | 41.1% | 42.4% | 37.7% – 47.2% |
| 41 | Jameis Winston | CLE | QB | 326 | 41.7% | 42.4% | 38.6% – 46.3% |
| 42 | Kenny Pickett◦ provisional | PHI | QB | 46 | 37.0% | 42.2% | 36.0% – 48.6% |
| 43 | Brandon Allen◦ provisional | SF | QB | 32 | 34.4% | 42.2% | 35.7% – 48.8% |
| 44 | Dak Prescott | DAL | QB | 307 | 41.4% | 42.2% | 38.3% – 46.1% |
| 45 | Mac Jones | JAX | QB | 279 | 41.2% | 42.1% | 38.1% – 46.2% |
| 46 | Cooper Rush | DAL | QB | 321 | 41.1% | 42.0% | 38.1% – 45.9% |
| 47 | C.J. Stroud | HOU | QB | 583 | 41.5% | 42.0% | 38.9% – 45.1% |
| 48 | Trey Lance◦ provisional | DAL | QB | 45 | 35.6% | 41.9% | 35.7% – 48.2% |
| 49 | Davis Mills◦ provisional | HOU | QB | 38 | 34.2% | 41.9% | 35.5% – 48.3% |
| 50 | Russell Wilson | PIT | QB | 368 | 41.0% | 41.8% | 38.2% – 45.5% |
| 51 | Desmond Ridder◦ provisional | LV | QB | 94 | 38.3% | 41.6% | 36.2% – 47.2% |
| 52 | Tim Boyle◦ provisional | NYG | QB | 51 | 35.3% | 41.6% | 35.5% – 47.8% |
| 53 | Bailey Zappe◦ provisional | CLE | QB | 32 | 31.3% | 41.5% | 35.0% – 48.1% |
| 54 | Jacoby Brissett | NE | QB | 179 | 39.1% | 41.2% | 36.5% – 45.9% |
| 55 | Skylar Thompson◦ provisional | MIA | QB | 39 | 30.8% | 41.0% | 34.6% – 47.4% |
| 56 | Tyler Huntley | MIA | QB | 149 | 37.6% | 40.6% | 35.7% – 45.5% |
| 57 | Bryce Young | CAR | QB | 414 | 39.1% | 40.3% | 36.8% – 43.8% |
| 58 | Jake Haener◦ provisional | NO | QB | 45 | 28.9% | 40.1% | 33.9% – 46.4% |
| 59 | Caleb Williams | CHI | QB | 636 | 38.8% | 39.7% | 36.8% – 42.6% |
| 60 | Spencer Rattler | NO | QB | 250 | 35.6% | 38.4% | 34.3% – 42.6% |
| 61 | Will Levis | TEN | QB | 344 | 35.8% | 38.0% | 34.3% – 41.7% |
| 62 | Anthony Richardson | IND | QB | 277 | 34.7% | 37.6% | 33.6% – 41.6% |
| 63 | Deshaun Watson | CLE | QB | 250 | 33.6% | 37.1% | 33.0% – 41.2% |
| 64 | Dorian Thompson-Robinson | CLE | QB | 126 | 27.8% | 35.8% | 30.9% – 40.9% |
How the shrinkage works
Two estimators
Rate stats (completion %, success rate, catch rate) use a beta-binomial model: a Beta(α, β) prior fit by marginal likelihood over each position group, then a Beta posterior per player. Per-play averages (EPA, CPOE, yards) use a normal-normal model with DerSimonian–Laird between-player variance. Both pull each player toward their group by exactly how thin their sample is.
Does it help? & the fine print
The trust panel’s numbers come from a leakage-free odd/even-week holdout: fit on odd weeks, predict even-week raw. Shrinkage lowers out-of-sample error for every stat. Priors are fit per season and per position (WR and TE separately), so a TE’s baseline isn’t a skill. Regular season only, 2016–2025. Rushing and receiving efficiency are heavily scheme-driven — read the wide bands as the honesty they are.
Source: nflverse play-by-play. Counted and computed deterministically — never modeled by a language model. Built by build_nfl_leaderboards.py (byte-reproducible).