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 · 2016
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
- 205
- 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 · 2016 · full board
| # | Player | Team | Pos | targets | Raw | Shrunk | 90% interval |
|---|---|---|---|---|---|---|---|
| 1 | Jordan Reed◦ provisional | WAS | TE | 90 | 62.2% | 55.2% | 50.4% – 59.9% |
| 2 | Cameron Brate◦ provisional | TB | TE | 81 | 61.7% | 54.8% | 50.0% – 59.6% |
| 3 | Travis Kelce◦ provisional | KC | TE | 118 | 59.3% | 54.7% | 50.2% – 59.3% |
| 4 | Jack Doyle◦ provisional | IND | TE | 75 | 61.3% | 54.6% | 49.7% – 59.4% |
| 5 | Greg Olsen◦ provisional | CAR | TE | 132 | 58.3% | 54.5% | 50.1% – 59.0% |
| 6 | Eric Ebron◦ provisional | DET | TE | 85 | 60.0% | 54.4% | 49.6% – 59.2% |
| 7 | Zach Ertz◦ provisional | PHI | TE | 106 | 58.5% | 54.3% | 49.6% – 58.9% |
| 8 | Dwayne Allen◦ provisional | IND | TE | 53 | 62.3% | 54.2% | 49.1% – 59.3% |
| 9 | Jimmy Graham◦ provisional | SEA | TE | 95 | 57.9% | 53.9% | 49.2% – 58.6% |
| 10 | Zach Miller◦ provisional | CHI | TE | 64 | 59.4% | 53.8% | 48.8% – 58.8% |
| 11 | Vernon Davis◦ provisional | WAS | TE | 59 | 59.3% | 53.7% | 48.6% – 58.7% |
| 12 | Daniel Brown◦ provisional | CHI | TE | 20 | 70.0% | 53.7% | 48.2% – 59.1% |
| 13 | Hunter Henry◦ provisional | LAC | TE | 54 | 59.3% | 53.6% | 48.5% – 58.7% |
| 14 | MarQueis Gray◦ provisional | MIA | TE | 17 | 70.6% | 53.5% | 48.0% – 59.0% |
| 15 | Rob Gronkowski◦ provisional | NE | TE | 38 | 60.5% | 53.4% | 48.1% – 58.6% |
| 16 | Will Tye◦ provisional | NYG | TE | 70 | 57.1% | 53.4% | 48.4% – 58.3% |
| 17 | Stephen Anderson◦ provisional | HOU | TE | 16 | 68.8% | 53.3% | 47.8% – 58.8% |
| 18 | Dion Sims◦ provisional | MIA | TE | 35 | 60.0% | 53.2% | 47.9% – 58.5% |
| 19 | Erik Swoope◦ provisional | IND | TE | 22 | 63.6% | 53.2% | 47.8% – 58.6% |
| 20 | Brent Celek◦ provisional | PHI | TE | 19 | 63.2% | 53.0% | 47.5% – 58.5% |
| 21 | Levine Toilolo◦ provisional | ATL | TE | 19 | 63.2% | 53.0% | 47.5% – 58.5% |
| 22 | Jerell Adams◦ provisional | NYG | TE | 21 | 61.9% | 53.0% | 47.5% – 58.4% |
| 23 | Tyler Kroft◦ provisional | CIN | TE | 12 | 66.7% | 52.9% | 47.3% – 58.4% |
| 24 | Darren Fells◦ provisional | ARI | TE | 18 | 61.1% | 52.8% | 47.3% – 58.3% |
| 25 | Austin Seferian-Jenkins◦ provisional | NYJ | TE | 20 | 60.0% | 52.8% | 47.3% – 58.2% |
| 26 | Ben Koyack◦ provisional | JAX | TE | 24 | 58.3% | 52.7% | 47.3% – 58.1% |
| 27 | Austin Hooper◦ provisional | ATL | TE | 28 | 57.1% | 52.7% | 47.3% – 58.1% |
| 28 | Mychal Rivera◦ provisional | LV | TE | 25 | 56.0% | 52.5% | 47.1% – 57.9% |
| 29 | Jacob Tamme◦ provisional | ATL | TE | 31 | 54.8% | 52.4% | 47.1% – 57.8% |
| 30 | Martellus Bennett◦ provisional | NE | TE | 73 | 53.4% | 52.4% | 47.5% – 57.4% |
| 31 | David Johnson◦ provisional | PIT | TE | 12 | 58.3% | 52.4% | 46.9% – 58.0% |
| 32 | Neal Sterling◦ provisional | JAX | TE | 16 | 56.3% | 52.4% | 46.9% – 57.9% |
| 33 | A.J. Derby◦ provisional | DEN | TE | 20 | 55.0% | 52.3% | 46.9% – 57.8% |
| 34 | Tyler Eifert◦ provisional | CIN | TE | 47 | 53.2% | 52.3% | 47.1% – 57.5% |
| 35 | Richard Rodgers◦ provisional | GB | TE | 47 | 53.2% | 52.3% | 47.1% – 57.5% |
| 36 | Brandon Myers◦ provisional | TB | TE | 15 | 53.3% | 52.2% | 46.6% – 57.7% |
| 37 | Nick O'Leary◦ provisional | BUF | TE | 15 | 53.3% | 52.2% | 46.6% – 57.7% |
| 38 | C.J. Uzomah◦ provisional | CIN | TE | 38 | 52.6% | 52.2% | 46.9% – 57.4% |
| 39 | Luke Willson◦ provisional | SEA | TE | 21 | 52.4% | 52.1% | 46.7% – 57.6% |
| 40 | Virgil Green◦ provisional | DEN | TE | 37 | 51.3% | 52.0% | 46.7% – 57.2% |
| 41 | Anthony Fasano◦ provisional | TEN | TE | 14 | 50.0% | 51.9% | 46.4% – 57.5% |
| 42 | Ryan Griffin◦ provisional | HOU | TE | 74 | 51.3% | 51.9% | 47.0% – 56.8% |
| 43 | Coby Fleener◦ provisional | NO | TE | 82 | 51.2% | 51.8% | 47.0% – 56.7% |
| 44 | Jermaine Gresham◦ provisional | ARI | TE | 61 | 50.8% | 51.8% | 46.8% – 56.8% |
| 45 | Demetrius Harris◦ provisional | KC | TE | 32 | 50.0% | 51.8% | 46.5% – 57.1% |
| 46 | John Phillips◦ provisional | NO | TE | 15 | 46.7% | 51.7% | 46.2% – 57.2% |
| 47 | Darren Waller◦ provisional | BAL | TE | 17 | 47.1% | 51.7% | 46.2% – 57.2% |
| 48 | Jeff Heuerman◦ provisional | DEN | TE | 17 | 47.1% | 51.7% | 46.2% – 57.2% |
| 49 | Ed Dickson◦ provisional | CAR | TE | 19 | 47.4% | 51.7% | 46.2% – 57.2% |
| 50 | Jason Witten◦ provisional | DAL | TE | 95 | 50.5% | 51.6% | 46.9% – 56.3% |
| 51 | Ladarius Green◦ provisional | PIT | TE | 35 | 48.6% | 51.6% | 46.3% – 56.9% |
| 52 | Seth DeValve◦ provisional | CLE | TE | 12 | 41.7% | 51.5% | 45.9% – 57.1% |
| 53 | Crockett Gillmore◦ provisional | BAL | TE | 14 | 42.9% | 51.5% | 45.9% – 57.0% |
| 54 | Antonio Gates◦ provisional | LAC | TE | 94 | 50.0% | 51.4% | 46.7% – 56.2% |
| 55 | Jesse James◦ provisional | PIT | TE | 61 | 49.2% | 51.4% | 46.4% – 56.5% |
| 56 | Garrett Celek◦ provisional | SF | TE | 50 | 48.0% | 51.3% | 46.1% – 56.4% |
| 57 | C.J. Fiedorowicz◦ provisional | HOU | TE | 89 | 49.4% | 51.3% | 46.5% – 56.1% |
| 58 | Jared Cook◦ provisional | GB | TE | 51 | 47.1% | 51.1% | 46.0% – 56.2% |
| 59 | Gary Barnidge◦ provisional | CLE | TE | 83 | 48.2% | 51.0% | 46.1% – 55.8% |
| 60 | Trey Burton◦ provisional | PHI | TE | 60 | 46.7% | 50.9% | 45.8% – 55.9% |
| 61 | Jordan Matthews◦ provisional | PHI | TE | 119 | 48.7% | 50.9% | 46.3% – 55.4% |
| 62 | Xavier Grimble◦ provisional | PIT | TE | 21 | 38.1% | 50.8% | 45.3% – 56.2% |
| 63 | Delanie Walker◦ provisional | TEN | TE | 102 | 48.0% | 50.7% | 46.1% – 55.4% |
| 64 | Josh Hill◦ provisional | NO | TE | 22 | 36.4% | 50.6% | 45.1% – 56.0% |
| 65 | Marcedes Lewis◦ provisional | JAX | TE | 30 | 40.0% | 50.5% | 45.2% – 55.9% |
| 66 | Julius Thomas◦ provisional | JAX | TE | 51 | 43.1% | 50.3% | 45.2% – 55.4% |
| 67 | Charles Clay◦ provisional | BUF | TE | 87 | 46.0% | 50.3% | 45.5% – 55.1% |
| 68 | Dennis Pitta◦ provisional | BAL | TE | 121 | 47.1% | 50.2% | 45.7% – 54.8% |
| 69 | Rhett Ellison◦ provisional | MIN | TE | 14 | 21.4% | 50.1% | 44.6% – 55.7% |
| 70 | Larry Donnell◦ provisional | NYG | TE | 22 | 31.8% | 50.1% | 44.7% – 55.6% |
| 71 | Kyle Rudolph◦ provisional | MIN | TE | 133 | 46.6% | 49.9% | 45.5% – 54.4% |
| 72 | Clive Walford◦ provisional | LV | TE | 52 | 40.4% | 49.7% | 44.6% – 54.8% |
| 73 | Vance McDonald◦ provisional | SF | TE | 45 | 37.8% | 49.5% | 44.3% – 54.7% |
| 74 | Lance Kendricks◦ provisional | LA | TE | 87 | 42.5% | 49.2% | 44.4% – 54.0% |
| 75 | Tyler Higbee◦ provisional | LA | TE | 29 | 13.8% | 47.3% | 42.0% – 52.7% |
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).