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.
Completion % · 2022
Completions per pass attempt. Depth-of-target confound: a checkdown offense completes more. Pair with CPOE.
- Beats raw by
- 15.6%
- lower out-of-sample error
- RMSE raw → shrunk
- 5.2% → 4.4%
- odd vs. even weeks
- Split-half reliability
- 0.42
- how repeatable the raw stat is
- Stabilizes at
- 324
- attempts 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.
Completion % · 2022 · full board
| # | Player | Team | Pos | attempts | Raw | Shrunk | 90% interval |
|---|---|---|---|---|---|---|---|
| 1 | Justin Herbert | LAC | QB | 738 | 64.6% | 63.1% | 60.6% – 65.5% |
| 2 | Tom Brady | TB | QB | 761 | 64.4% | 62.9% | 60.5% – 65.3% |
| 3 | Patrick Mahomes | KC | QB | 677 | 64.3% | 62.7% | 60.2% – 65.2% |
| 4 | Geno Smith | SEA | QB | 620 | 64.3% | 62.7% | 60.1% – 65.3% |
| 5 | Joe Burrow | CIN | QB | 652 | 63.5% | 62.2% | 59.6% – 64.7% |
| 6 | Trevor Lawrence | JAX | QB | 617 | 62.7% | 61.6% | 59.0% – 64.2% |
| 7 | Jimmy Garoppolo | SF | QB | 326 | 63.5% | 61.5% | 58.4% – 64.6% |
| 8 | Dak Prescott | DAL | QB | 414 | 63.0% | 61.5% | 58.6% – 64.4% |
| 9 | Jared Goff | DET | QB | 611 | 62.5% | 61.5% | 58.9% – 64.1% |
| 10 | Nick Mullens◦ provisional | MIN | QB | 25 | 84.0% | 61.3% | 57.0% – 65.6% |
| 11 | Bailey Zappe◦ provisional | NE | QB | 98 | 66.3% | 61.1% | 57.2% – 65.0% |
| 12 | Andy Dalton | NO | QB | 404 | 62.4% | 61.1% | 58.1% – 64.1% |
| 13 | Matthew Stafford | LA | QB | 332 | 62.1% | 60.8% | 57.7% – 63.9% |
| 14 | Brock Purdy◦ provisional | SF | QB | 181 | 63.0% | 60.8% | 57.2% – 64.3% |
| 15 | Matt Ryan | IND | QB | 503 | 61.4% | 60.7% | 57.9% – 63.5% |
| 16 | Kyler Murray | ARI | QB | 421 | 61.5% | 60.7% | 57.7% – 63.6% |
| 17 | Kirk Cousins | MIN | QB | 693 | 61.2% | 60.7% | 58.1% – 63.2% |
| 18 | Jalen Hurts | PHI | QB | 499 | 61.3% | 60.6% | 57.8% – 63.4% |
| 19 | Tua Tagovailoa | MIA | QB | 422 | 61.4% | 60.6% | 57.6% – 63.5% |
| 20 | Daniel Jones | NYG | QB | 519 | 61.1% | 60.5% | 57.7% – 63.2% |
| 21 | Tyler Huntley◦ provisional | BAL | QB | 119 | 63.0% | 60.5% | 56.6% – 64.3% |
| 22 | Colt McCoy◦ provisional | ARI | QB | 145 | 62.1% | 60.3% | 56.6% – 64.0% |
| 23 | Aaron Rodgers | GB | QB | 577 | 60.7% | 60.3% | 57.6% – 62.9% |
| 24 | Mitchell Trubisky◦ provisional | PIT | QB | 192 | 60.9% | 60.1% | 56.5% – 63.6% |
| 25 | Mac Jones | NE | QB | 480 | 60.0% | 59.8% | 57.0% – 62.6% |
| 26 | David Blough◦ provisional | ARI | QB | 63 | 60.3% | 59.7% | 55.5% – 63.7% |
| 27 | Josh Allen | BUF | QB | 601 | 59.7% | 59.7% | 57.0% – 62.3% |
| 28 | Jacoby Brissett | CLE | QB | 395 | 59.8% | 59.7% | 56.6% – 62.6% |
| 29 | Jarrett Stidham◦ provisional | LV | QB | 90 | 58.9% | 59.4% | 55.4% – 63.3% |
| 30 | Desmond Ridder◦ provisional | ATL | QB | 124 | 58.9% | 59.4% | 55.5% – 63.2% |
| 31 | Ryan Tannehill | TEN | QB | 359 | 59.1% | 59.3% | 56.2% – 62.4% |
| 32 | Davis Webb◦ provisional | NYG | QB | 41 | 56.1% | 59.2% | 54.9% – 63.4% |
| 33 | Nathan Peterman◦ provisional | CHI | QB | 26 | 53.8% | 59.1% | 54.8% – 63.4% |
| 34 | Trevor Siemian◦ provisional | CHI | QB | 28 | 53.6% | 59.1% | 54.7% – 63.4% |
| 35 | Kenny Pickett | PIT | QB | 418 | 58.6% | 59.0% | 56.0% – 62.0% |
| 36 | Jameis Winston◦ provisional | NO | QB | 127 | 57.5% | 59.0% | 55.1% – 62.8% |
| 37 | Teddy Bridgewater◦ provisional | MIA | QB | 87 | 56.3% | 58.9% | 54.9% – 62.8% |
| 38 | Taylor Heinicke◦ provisional | WAS | QB | 278 | 57.9% | 58.8% | 55.5% – 62.1% |
| 39 | John Wolford◦ provisional | LA | QB | 69 | 55.1% | 58.8% | 54.7% – 62.8% |
| 40 | PJ Walker◦ provisional | CAR | QB | 113 | 55.8% | 58.6% | 54.7% – 62.4% |
| 41 | Brett Rypien◦ provisional | DEN | QB | 96 | 55.2% | 58.6% | 54.6% – 62.5% |
| 42 | Joshua Dobbs◦ provisional | TEN | QB | 74 | 54.0% | 58.5% | 54.4% – 62.6% |
| 43 | Sam Ehlinger◦ provisional | IND | QB | 115 | 55.6% | 58.5% | 54.6% – 62.4% |
| 44 | Lamar Jackson | BAL | QB | 353 | 57.5% | 58.5% | 55.4% – 61.6% |
| 45 | Kyle Allen◦ provisional | HOU | QB | 85 | 54.1% | 58.4% | 54.4% – 62.4% |
| 46 | Gardner Minshew◦ provisional | PHI | QB | 82 | 53.7% | 58.4% | 54.3% – 62.4% |
| 47 | Nick Foles◦ provisional | IND | QB | 50 | 50.0% | 58.3% | 54.1% – 62.4% |
| 48 | Mike White◦ provisional | NYJ | QB | 184 | 56.0% | 58.3% | 54.6% – 61.8% |
| 49 | Trey Lance◦ provisional | SF | QB | 33 | 45.5% | 58.3% | 53.9% – 62.5% |
| 50 | Bryce Perkins◦ provisional | LA | QB | 40 | 47.5% | 58.2% | 53.9% – 62.5% |
| 51 | Cooper Rush◦ provisional | DAL | QB | 169 | 55.6% | 58.2% | 54.5% – 61.8% |
| 52 | Derek Carr | LV | QB | 532 | 57.3% | 58.2% | 55.4% – 60.9% |
| 53 | Skylar Thompson◦ provisional | MIA | QB | 111 | 54.0% | 58.1% | 54.2% – 62.0% |
| 54 | Sam Darnold◦ provisional | CAR | QB | 150 | 54.7% | 58.0% | 54.3% – 61.7% |
| 55 | Davis Mills | HOU | QB | 512 | 57.0% | 58.0% | 55.2% – 60.8% |
| 56 | Trace McSorley◦ provisional | ARI | QB | 87 | 51.7% | 57.9% | 53.9% – 61.9% |
| 57 | Joe Flacco◦ provisional | NYJ | QB | 201 | 54.7% | 57.7% | 54.1% – 61.2% |
| 58 | Carson Wentz◦ provisional | WAS | QB | 309 | 55.7% | 57.6% | 54.4% – 60.9% |
| 59 | Marcus Mariota | ATL | QB | 330 | 55.8% | 57.6% | 54.4% – 60.8% |
| 60 | Anthony Brown◦ provisional | BAL | QB | 54 | 40.7% | 56.9% | 52.7% – 61.0% |
| 61 | Deshaun Watson◦ provisional | CLE | QB | 190 | 52.1% | 56.8% | 53.2% – 60.4% |
| 62 | Malik Willis◦ provisional | TEN | QB | 71 | 43.7% | 56.7% | 52.6% – 60.8% |
| 63 | Baker Mayfield | LA | QB | 372 | 54.0% | 56.6% | 53.5% – 59.7% |
| 64 | Russell Wilson | DEN | QB | 539 | 54.2% | 56.2% | 53.4% – 59.0% |
| 65 | Zach Wilson◦ provisional | NYJ | QB | 265 | 49.8% | 55.2% | 51.8% – 58.5% |
| 66 | Justin Fields | CHI | QB | 377 | 50.9% | 54.9% | 51.8% – 58.0% |
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).