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.
Catch rate · 2017
Receptions per target, shrunk within position (WR vs TE separately). Depth-of-target confound; catchable-target rate not available here.
- Beats raw by
- 18.3%
- lower out-of-sample error
- RMSE raw → shrunk
- 11.1% → 9.1%
- odd vs. even weeks
- Split-half reliability
- 0.35
- how repeatable the raw stat is
- Stabilizes at
- 97
- 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.
Catch rate · 2017 · full board
| # | Player | Team | Pos | targets | Raw | Shrunk | 90% interval |
|---|---|---|---|---|---|---|---|
| 1 | Benjamin Watson◦ provisional | BAL | TE | 80 | 76.3% | 69.3% | 63.5% – 74.9% |
| 2 | Jack Doyle | IND | TE | 109 | 73.4% | 68.8% | 63.4% – 74.0% |
| 3 | Austin Hooper◦ provisional | ATL | TE | 65 | 75.4% | 68.3% | 62.2% – 74.2% |
| 4 | Jason Witten◦ provisional | DAL | TE | 87 | 72.4% | 67.8% | 62.0% – 73.3% |
| 5 | Jordan Reed◦ provisional | WAS | TE | 35 | 77.1% | 67.2% | 60.3% – 73.7% |
| 6 | Hunter Henry◦ provisional | LAC | TE | 62 | 72.6% | 67.1% | 60.9% – 73.1% |
| 7 | Kyle Rudolph◦ provisional | MIN | TE | 81 | 70.4% | 66.7% | 60.8% – 72.4% |
| 8 | Maxx Williams◦ provisional | BAL | TE | 18 | 83.3% | 66.7% | 59.3% – 73.7% |
| 9 | Nick Boyle◦ provisional | BAL | TE | 38 | 73.7% | 66.4% | 59.6% – 72.9% |
| 10 | Rhett Ellison◦ provisional | NYG | TE | 32 | 75.0% | 66.4% | 59.5% – 73.1% |
| 11 | Levine Toilolo◦ provisional | ATL | TE | 14 | 85.7% | 66.4% | 58.9% – 73.5% |
| 12 | Adam Shaheen◦ provisional | CHI | TE | 14 | 85.7% | 66.4% | 58.9% – 73.5% |
| 13 | Jermaine Gresham◦ provisional | ARI | TE | 46 | 71.7% | 66.2% | 59.6% – 72.6% |
| 14 | Travis Kelce | KC | TE | 122 | 68.0% | 66.1% | 60.7% – 71.2% |
| 15 | Coby Fleener◦ provisional | NO | TE | 30 | 73.3% | 65.9% | 58.9% – 72.6% |
| 16 | Nick Vannett◦ provisional | SEA | TE | 15 | 80.0% | 65.8% | 58.3% – 73.0% |
| 17 | David Morgan◦ provisional | MIN | TE | 12 | 83.3% | 65.8% | 58.1% – 73.0% |
| 18 | Trey Burton◦ provisional | PHI | TE | 32 | 71.9% | 65.6% | 58.7% – 72.4% |
| 19 | Brandon Williams◦ provisional | IND | TE | 17 | 76.5% | 65.5% | 58.1% – 72.6% |
| 20 | Jesse James◦ provisional | PIT | TE | 63 | 68.3% | 65.4% | 59.1% – 71.5% |
| 21 | George Kittle◦ provisional | SF | TE | 63 | 68.3% | 65.4% | 59.1% – 71.5% |
| 22 | Austin Seferian-Jenkins◦ provisional | NYJ | TE | 74 | 67.6% | 65.3% | 59.2% – 71.2% |
| 23 | Josh Hill◦ provisional | NO | TE | 22 | 72.7% | 65.3% | 58.0% – 72.3% |
| 24 | Delanie Walker | TEN | TE | 111 | 66.7% | 65.2% | 59.7% – 70.6% |
| 25 | Zach Ertz | PHI | TE | 111 | 66.7% | 65.2% | 59.7% – 70.6% |
| 26 | Tyler Kroft◦ provisional | CIN | TE | 62 | 67.7% | 65.2% | 58.9% – 71.3% |
| 27 | Anthony Fasano◦ provisional | MIA | TE | 16 | 75.0% | 65.2% | 57.7% – 72.4% |
| 28 | Jordan Matthews◦ provisional | BUF | TE | 36 | 69.4% | 65.2% | 58.3% – 71.8% |
| 29 | Martellus Bennett◦ provisional | NE | TE | 44 | 68.2% | 65.0% | 58.3% – 71.5% |
| 30 | Nick O'Leary◦ provisional | BUF | TE | 32 | 68.8% | 64.9% | 57.8% – 71.6% |
| 31 | Charles Clay◦ provisional | BUF | TE | 74 | 66.2% | 64.7% | 58.6% – 70.6% |
| 32 | Julius Thomas◦ provisional | MIA | TE | 62 | 66.1% | 64.6% | 58.3% – 70.7% |
| 33 | O.J. Howard◦ provisional | TB | TE | 39 | 66.7% | 64.5% | 57.6% – 71.1% |
| 34 | Luke Willson◦ provisional | SEA | TE | 22 | 68.2% | 64.4% | 57.1% – 71.5% |
| 35 | Rob Gronkowski | NE | TE | 106 | 65.1% | 64.4% | 58.8% – 69.8% |
| 36 | Niles Paul◦ provisional | WAS | TE | 19 | 68.4% | 64.4% | 57.0% – 71.5% |
| 37 | Clive Walford◦ provisional | LV | TE | 13 | 69.2% | 64.3% | 56.6% – 71.6% |
| 38 | C.J. Uzomah◦ provisional | CIN | TE | 15 | 66.7% | 64.0% | 56.4% – 71.3% |
| 39 | Darren Fells◦ provisional | DET | TE | 26 | 65.4% | 64.0% | 56.7% – 70.9% |
| 40 | Daniel Brown◦ provisional | CHI | TE | 20 | 65.0% | 63.8% | 56.4% – 71.0% |
| 41 | Virgil Green◦ provisional | DEN | TE | 22 | 63.6% | 63.6% | 56.2% – 70.7% |
| 42 | Garrett Celek◦ provisional | SF | TE | 33 | 63.6% | 63.6% | 56.6% – 70.4% |
| 43 | C.J. Fiedorowicz◦ provisional | HOU | TE | 22 | 63.6% | 63.6% | 56.2% – 70.7% |
| 44 | Richard Rodgers◦ provisional | GB | TE | 19 | 63.2% | 63.5% | 56.0% – 70.7% |
| 45 | Austin Traylor◦ provisional | DEN | TE | 13 | 61.5% | 63.3% | 55.7% – 70.7% |
| 46 | Jared Cook◦ provisional | LV | TE | 86 | 62.8% | 63.2% | 57.3% – 69.0% |
| 47 | Vernon Davis◦ provisional | WAS | TE | 69 | 62.3% | 63.0% | 56.8% – 69.1% |
| 48 | Cameron Brate◦ provisional | TB | TE | 77 | 62.3% | 63.0% | 56.9% – 69.0% |
| 49 | Ed Dickson◦ provisional | CAR | TE | 49 | 61.2% | 62.8% | 56.1% – 69.3% |
| 50 | Jonnu Smith◦ provisional | TEN | TE | 30 | 60.0% | 62.7% | 55.6% – 69.7% |
| 51 | Eric Ebron◦ provisional | DET | TE | 86 | 61.6% | 62.7% | 56.7% – 68.5% |
| 52 | Vance McDonald◦ provisional | PIT | TE | 24 | 58.3% | 62.5% | 55.2% – 69.6% |
| 53 | James O'Shaughnessy◦ provisional | JAX | TE | 24 | 58.3% | 62.5% | 55.2% – 69.6% |
| 54 | Zach Miller◦ provisional | CHI | TE | 35 | 57.1% | 61.9% | 54.8% – 68.7% |
| 55 | Brent Celek◦ provisional | PHI | TE | 24 | 54.2% | 61.7% | 54.4% – 68.8% |
| 56 | Jeff Heuerman◦ provisional | DEN | TE | 18 | 50.0% | 61.5% | 53.9% – 68.8% |
| 57 | Antonio Gates◦ provisional | LAC | TE | 53 | 56.6% | 61.1% | 54.5% – 67.6% |
| 58 | Seth DeValve◦ provisional | CLE | TE | 58 | 56.9% | 61.1% | 54.6% – 67.4% |
| 59 | Tyler Higbee◦ provisional | LA | TE | 45 | 55.6% | 61.0% | 54.2% – 67.7% |
| 60 | Jimmy Graham | SEA | TE | 98 | 58.2% | 60.9% | 55.1% – 66.5% |
| 61 | Dion Sims◦ provisional | CHI | TE | 29 | 51.7% | 60.9% | 53.6% – 67.9% |
| 62 | Troy Niklas◦ provisional | ARI | TE | 23 | 47.8% | 60.6% | 53.1% – 67.8% |
| 63 | Lance Kendricks◦ provisional | GB | TE | 35 | 51.4% | 60.4% | 53.3% – 67.3% |
| 64 | A.J. Derby◦ provisional | MIA | TE | 40 | 52.5% | 60.3% | 53.4% – 67.1% |
| 65 | Ross Travis◦ provisional | IND | TE | 17 | 41.2% | 60.2% | 52.6% – 67.6% |
| 66 | Dwayne Allen◦ provisional | NE | TE | 22 | 45.5% | 60.2% | 52.8% – 67.5% |
| 67 | Ryan Griffin◦ provisional | HOU | TE | 27 | 48.1% | 60.2% | 52.9% – 67.3% |
| 68 | Gerald Everett◦ provisional | LA | TE | 32 | 50.0% | 60.2% | 53.1% – 67.2% |
| 69 | Demetrius Harris◦ provisional | KC | TE | 36 | 50.0% | 59.9% | 52.8% – 66.8% |
| 70 | David Njoku◦ provisional | CLE | TE | 60 | 53.3% | 59.7% | 53.2% – 66.0% |
| 71 | Evan Engram | NYG | TE | 115 | 55.6% | 59.3% | 53.7% – 64.8% |
| 72 | Ricky Seals-Jones◦ provisional | ARI | TE | 28 | 42.9% | 58.9% | 51.6% – 66.1% |
| 73 | Darrell Daniels◦ provisional | IND | TE | 13 | 23.1% | 58.8% | 51.0% – 66.4% |
| 74 | Marcedes Lewis◦ provisional | JAX | TE | 49 | 49.0% | 58.7% | 51.9% – 65.3% |
| 75 | Greg Olsen◦ provisional | CAR | TE | 38 | 44.7% | 58.3% | 51.2% – 65.2% |
| 76 | Stephen Anderson◦ provisional | HOU | TE | 52 | 48.1% | 58.2% | 51.5% – 64.7% |
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).