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.
Target success rate · 2018
Share of targets with positive EPA, shrunk within position.
- Beats raw by
- 21.9%
- lower out-of-sample error
- RMSE raw → shrunk
- 11.7% → 9.1%
- odd vs. even weeks
- Split-half reliability
- 0.19
- how repeatable the raw stat is
- Stabilizes at
- 4737
- targets 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.
Target success rate · 2018 · full board
| # | Player | Team | Pos | targets | Raw | Shrunk | 90% interval |
|---|---|---|---|---|---|---|---|
| 1 | Rob Gronkowski◦ provisional | NE | TE | 72 | 63.9% | 54.0% | 52.9% – 55.2% |
| 2 | Travis Kelce◦ provisional | KC | TE | 151 | 57.6% | 54.0% | 52.8% – 55.2% |
| 3 | Levine Toilolo◦ provisional | DET | TE | 24 | 75.0% | 54.0% | 52.8% – 55.2% |
| 4 | O.J. Howard◦ provisional | TB | TE | 48 | 64.6% | 54.0% | 52.8% – 55.2% |
| 5 | Mark Andrews◦ provisional | BAL | TE | 50 | 64.0% | 54.0% | 52.8% – 55.2% |
| 6 | Kyle Rudolph◦ provisional | MIN | TE | 84 | 59.5% | 54.0% | 52.8% – 55.2% |
| 7 | Jesse James◦ provisional | PIT | TE | 40 | 65.0% | 54.0% | 52.8% – 55.2% |
| 8 | Zach Ertz◦ provisional | PHI | TE | 156 | 56.4% | 54.0% | 52.8% – 55.1% |
| 9 | Anthony Firkser◦ provisional | TEN | TE | 20 | 75.0% | 54.0% | 52.8% – 55.2% |
| 10 | Greg Olsen◦ provisional | CAR | TE | 38 | 63.2% | 54.0% | 52.8% – 55.1% |
| 11 | Darren Fells◦ provisional | CLE | TE | 12 | 83.3% | 54.0% | 52.8% – 55.2% |
| 12 | Benjamin Watson◦ provisional | NO | TE | 46 | 60.9% | 54.0% | 52.8% – 55.1% |
| 13 | Vernon Davis◦ provisional | WAS | TE | 37 | 62.2% | 54.0% | 52.8% – 55.1% |
| 14 | Jared Cook◦ provisional | LV | TE | 101 | 56.4% | 53.9% | 52.8% – 55.1% |
| 15 | Lance Kendricks◦ provisional | GB | TE | 25 | 64.0% | 53.9% | 52.8% – 55.1% |
| 16 | Tyler Eifert◦ provisional | CIN | TE | 19 | 68.4% | 53.9% | 52.8% – 55.1% |
| 17 | Maxx Williams◦ provisional | BAL | TE | 17 | 70.6% | 53.9% | 52.8% – 55.1% |
| 18 | C.J. Uzomah◦ provisional | CIN | TE | 64 | 57.8% | 53.9% | 52.8% – 55.1% |
| 19 | Tyler Higbee◦ provisional | LA | TE | 34 | 61.8% | 53.9% | 52.8% – 55.1% |
| 20 | George Kittle◦ provisional | SF | TE | 138 | 55.8% | 53.9% | 52.8% – 55.1% |
| 21 | Blake Jarwin◦ provisional | DAL | TE | 36 | 61.1% | 53.9% | 52.8% – 55.1% |
| 22 | Rhett Ellison◦ provisional | NYG | TE | 35 | 60.0% | 53.9% | 52.8% – 55.1% |
| 23 | Austin Hooper◦ provisional | ATL | TE | 90 | 55.6% | 53.9% | 52.8% – 55.1% |
| 24 | Jonnu Smith◦ provisional | TEN | TE | 30 | 60.0% | 53.9% | 52.8% – 55.1% |
| 25 | Gerald Everett◦ provisional | LA | TE | 51 | 56.9% | 53.9% | 52.7% – 55.1% |
| 26 | Ed Dickson◦ provisional | SEA | TE | 13 | 61.5% | 53.9% | 52.7% – 55.1% |
| 27 | Jack Doyle◦ provisional | IND | TE | 33 | 57.6% | 53.9% | 52.7% – 55.1% |
| 28 | Vance McDonald◦ provisional | PIT | TE | 72 | 55.6% | 53.9% | 52.7% – 55.1% |
| 29 | Matt LaCosse◦ provisional | DEN | TE | 37 | 56.8% | 53.9% | 52.7% – 55.1% |
| 30 | Dallas Goedert◦ provisional | PHI | TE | 44 | 56.8% | 53.9% | 52.7% – 55.1% |
| 31 | Derek Carrier◦ provisional | LV | TE | 12 | 58.3% | 53.9% | 52.7% – 55.1% |
| 32 | Trey Burton◦ provisional | CHI | TE | 77 | 54.5% | 53.9% | 52.7% – 55.1% |
| 33 | Jordan Matthews◦ provisional | PHI | TE | 28 | 57.1% | 53.9% | 52.7% – 55.1% |
| 34 | Dalton Schultz◦ provisional | DAL | TE | 17 | 58.8% | 53.9% | 52.7% – 55.1% |
| 35 | Virgil Green◦ provisional | LAC | TE | 27 | 55.6% | 53.9% | 52.7% – 55.1% |
| 36 | Jermaine Gresham◦ provisional | ARI | TE | 12 | 50.0% | 53.9% | 52.7% – 55.1% |
| 37 | Luke Stocker◦ provisional | TEN | TE | 21 | 52.4% | 53.9% | 52.7% – 55.1% |
| 38 | Luke Willson◦ provisional | DET | TE | 19 | 52.6% | 53.9% | 52.7% – 55.1% |
| 39 | Ryan Griffin◦ provisional | HOU | TE | 43 | 53.5% | 53.9% | 52.7% – 55.1% |
| 40 | Dan Arnold◦ provisional | NO | TE | 19 | 52.6% | 53.9% | 52.7% – 55.1% |
| 41 | Ian Thomas◦ provisional | CAR | TE | 49 | 53.1% | 53.9% | 52.7% – 55.1% |
| 42 | Chris Herndon◦ provisional | NYJ | TE | 56 | 53.6% | 53.9% | 52.7% – 55.1% |
| 43 | Josh Hill◦ provisional | NO | TE | 24 | 50.0% | 53.9% | 52.7% – 55.1% |
| 44 | Scott Simonson◦ provisional | NYG | TE | 14 | 50.0% | 53.9% | 52.7% – 55.1% |
| 45 | Nick Boyle◦ provisional | BAL | TE | 37 | 51.3% | 53.9% | 52.7% – 55.1% |
| 46 | Niles Paul◦ provisional | JAX | TE | 13 | 46.2% | 53.9% | 52.7% – 55.1% |
| 47 | Nick Vannett◦ provisional | SEA | TE | 43 | 51.2% | 53.9% | 52.7% – 55.1% |
| 48 | Mo Alie-Cox◦ provisional | IND | TE | 13 | 46.2% | 53.9% | 52.7% – 55.1% |
| 49 | Jake Butt◦ provisional | DEN | TE | 13 | 46.2% | 53.9% | 52.7% – 55.1% |
| 50 | Eric Tomlinson◦ provisional | NYJ | TE | 14 | 42.9% | 53.9% | 52.7% – 55.0% |
| 51 | Jordan Leggett◦ provisional | NYJ | TE | 25 | 48.0% | 53.9% | 52.7% – 55.0% |
| 52 | Will Dissly◦ provisional | SEA | TE | 14 | 42.9% | 53.9% | 52.7% – 55.0% |
| 53 | Jordan Akins◦ provisional | HOU | TE | 25 | 48.0% | 53.9% | 52.7% – 55.0% |
| 54 | Logan Thomas◦ provisional | BUF | TE | 17 | 41.2% | 53.8% | 52.7% – 55.0% |
| 55 | Geoff Swaim◦ provisional | DAL | TE | 32 | 46.9% | 53.8% | 52.7% – 55.0% |
| 56 | Antonio Gates◦ provisional | LAC | TE | 48 | 47.9% | 53.8% | 52.6% – 55.0% |
| 57 | Demetrius Harris◦ provisional | KC | TE | 25 | 44.0% | 53.8% | 52.6% – 55.0% |
| 58 | James O'Shaughnessy◦ provisional | JAX | TE | 38 | 47.4% | 53.8% | 52.7% – 55.0% |
| 59 | Hayden Hurst◦ provisional | BAL | TE | 23 | 43.5% | 53.8% | 52.7% – 55.0% |
| 60 | Austin Seferian-Jenkins◦ provisional | JAX | TE | 19 | 36.8% | 53.8% | 52.6% – 55.0% |
| 61 | Mike Gesicki◦ provisional | MIA | TE | 32 | 43.8% | 53.8% | 52.6% – 55.0% |
| 62 | Cameron Brate◦ provisional | TB | TE | 49 | 46.9% | 53.8% | 52.6% – 55.0% |
| 63 | Eric Ebron◦ provisional | IND | TE | 110 | 50.0% | 53.8% | 52.6% – 55.0% |
| 64 | Jeff Heuerman◦ provisional | DEN | TE | 48 | 45.8% | 53.8% | 52.6% – 55.0% |
| 65 | Jason Croom◦ provisional | BUF | TE | 35 | 42.9% | 53.8% | 52.6% – 55.0% |
| 66 | Charles Clay◦ provisional | BUF | TE | 36 | 41.7% | 53.8% | 52.6% – 55.0% |
| 67 | Jordan Reed◦ provisional | WAS | TE | 84 | 47.6% | 53.8% | 52.6% – 55.0% |
| 68 | Michael Roberts◦ provisional | DET | TE | 20 | 30.0% | 53.8% | 52.6% – 55.0% |
| 69 | Evan Engram◦ provisional | NYG | TE | 65 | 46.2% | 53.8% | 52.6% – 55.0% |
| 70 | David Njoku◦ provisional | CLE | TE | 88 | 46.6% | 53.8% | 52.6% – 54.9% |
| 71 | Jimmy Graham◦ provisional | GB | TE | 89 | 46.1% | 53.8% | 52.6% – 54.9% |
| 72 | Ricky Seals-Jones◦ provisional | ARI | TE | 69 | 36.2% | 53.6% | 52.5% – 54.8% |
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).