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 · 2023
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
- 334
- 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 · 2023 · full board
| # | Player | Team | Pos | targets | Raw | Shrunk | 90% interval |
|---|---|---|---|---|---|---|---|
| 1 | George Kittle◦ provisional | SF | TE | 90 | 63.3% | 55.4% | 51.4% – 59.4% |
| 2 | Durham Smythe◦ provisional | MIA | TE | 43 | 69.8% | 55.2% | 51.0% – 59.4% |
| 3 | Tucker Kraft◦ provisional | GB | TE | 40 | 70.0% | 55.1% | 50.8% – 59.3% |
| 4 | Cole Kmet◦ provisional | CHI | TE | 90 | 61.1% | 55.0% | 51.0% – 58.9% |
| 5 | Travis Kelce◦ provisional | KC | TE | 121 | 59.5% | 54.9% | 51.1% – 58.8% |
| 6 | Mark Andrews◦ provisional | BAL | TE | 61 | 63.9% | 54.9% | 50.8% – 59.1% |
| 7 | Hunter Henry◦ provisional | NE | TE | 61 | 62.3% | 54.7% | 50.6% – 58.8% |
| 8 | Brevin Jordan◦ provisional | HOU | TE | 21 | 76.2% | 54.7% | 50.3% – 59.0% |
| 9 | Sam LaPorta◦ provisional | DET | TE | 122 | 58.2% | 54.6% | 50.8% – 58.4% |
| 10 | Isaiah Likely◦ provisional | BAL | TE | 40 | 65.0% | 54.6% | 50.3% – 58.8% |
| 11 | Taysom Hill◦ provisional | NO | TE | 40 | 62.5% | 54.3% | 50.0% – 58.5% |
| 12 | Dallas Goedert◦ provisional | PHI | TE | 83 | 57.8% | 54.2% | 50.2% – 58.2% |
| 13 | Tanner Hudson◦ provisional | CIN | TE | 50 | 60.0% | 54.2% | 50.0% – 58.3% |
| 14 | Donald Parham◦ provisional | LAC | TE | 41 | 61.0% | 54.1% | 49.9% – 58.4% |
| 15 | Dalton Kincaid◦ provisional | BUF | TE | 91 | 57.1% | 54.1% | 50.1% – 58.1% |
| 16 | Austin Hooper◦ provisional | LV | TE | 32 | 62.5% | 54.1% | 49.8% – 58.4% |
| 17 | Luke Farrell◦ provisional | JAX | TE | 15 | 66.7% | 53.9% | 49.5% – 58.3% |
| 18 | Logan Thomas◦ provisional | WAS | TE | 79 | 55.7% | 53.8% | 49.7% – 57.8% |
| 19 | Jake Ferguson◦ provisional | DAL | TE | 102 | 54.9% | 53.7% | 49.7% – 57.6% |
| 20 | Pharaoh Brown◦ provisional | NE | TE | 15 | 60.0% | 53.6% | 49.2% – 58.0% |
| 21 | Josh Whyle◦ provisional | TEN | TE | 15 | 60.0% | 53.6% | 49.2% – 58.0% |
| 22 | MyCole Pruitt◦ provisional | ATL | TE | 12 | 58.3% | 53.5% | 49.1% – 57.9% |
| 23 | Darren Waller◦ provisional | NYG | TE | 74 | 54.0% | 53.4% | 49.4% – 57.5% |
| 24 | Pat Freiermuth◦ provisional | PIT | TE | 48 | 54.2% | 53.4% | 49.2% – 57.6% |
| 25 | Jeremy Ruckert◦ provisional | NYJ | TE | 22 | 54.5% | 53.4% | 49.0% – 57.7% |
| 26 | Evan Engram◦ provisional | JAX | TE | 144 | 53.5% | 53.4% | 49.6% – 57.1% |
| 27 | Will Mallory◦ provisional | IND | TE | 26 | 53.8% | 53.3% | 49.0% – 57.7% |
| 28 | Michael Mayer◦ provisional | LV | TE | 41 | 53.7% | 53.3% | 49.1% – 57.6% |
| 29 | Noah Fant◦ provisional | SEA | TE | 43 | 53.5% | 53.3% | 49.1% – 57.5% |
| 30 | Kyle Pitts◦ provisional | ATL | TE | 90 | 53.3% | 53.3% | 49.3% – 57.3% |
| 31 | Elijah Higgins◦ provisional | ARI | TE | 19 | 52.6% | 53.3% | 48.9% – 57.6% |
| 32 | Cade Otton◦ provisional | TB | TE | 68 | 52.9% | 53.2% | 49.1% – 57.3% |
| 33 | Foster Moreau◦ provisional | NO | TE | 25 | 52.0% | 53.2% | 48.9% – 57.5% |
| 34 | C.J. Uzomah◦ provisional | NYJ | TE | 12 | 50.0% | 53.2% | 48.8% – 57.6% |
| 35 | Mitchell Wilcox◦ provisional | CIN | TE | 12 | 50.0% | 53.2% | 48.8% – 57.6% |
| 36 | Mike Gesicki◦ provisional | NE | TE | 46 | 52.2% | 53.2% | 48.9% – 57.4% |
| 37 | Brock Wright◦ provisional | DET | TE | 14 | 50.0% | 53.2% | 48.8% – 57.6% |
| 38 | Lucas Krull◦ provisional | DEN | TE | 14 | 50.0% | 53.2% | 48.8% – 57.6% |
| 39 | Dawson Knox◦ provisional | BUF | TE | 37 | 51.3% | 53.1% | 48.8% – 57.4% |
| 40 | Will Dissly◦ provisional | SEA | TE | 22 | 50.0% | 53.1% | 48.7% – 57.4% |
| 41 | Noah Gray◦ provisional | KC | TE | 41 | 51.2% | 53.1% | 48.8% – 57.3% |
| 42 | Josh Oliver◦ provisional | MIN | TE | 28 | 50.0% | 53.0% | 48.7% – 57.4% |
| 43 | Tommy Tremble◦ provisional | CAR | TE | 32 | 50.0% | 53.0% | 48.7% – 57.3% |
| 44 | Cole Turner◦ provisional | WAS | TE | 15 | 46.7% | 53.0% | 48.6% – 57.4% |
| 45 | Robert Tonyan◦ provisional | CHI | TE | 17 | 47.1% | 53.0% | 48.6% – 57.4% |
| 46 | Colby Parkinson◦ provisional | SEA | TE | 34 | 50.0% | 53.0% | 48.7% – 57.3% |
| 47 | Trey McBride◦ provisional | ARI | TE | 106 | 51.9% | 53.0% | 49.0% – 56.9% |
| 48 | Johnny Mundt◦ provisional | MIN | TE | 23 | 47.8% | 52.9% | 48.6% – 57.3% |
| 49 | Drew Sample◦ provisional | CIN | TE | 27 | 48.1% | 52.9% | 48.6% – 57.2% |
| 50 | Luke Musgrave◦ provisional | GB | TE | 46 | 50.0% | 52.9% | 48.7% – 57.1% |
| 51 | Harrison Bryant◦ provisional | CLE | TE | 20 | 45.0% | 52.8% | 48.5% – 57.2% |
| 52 | Daniel Bellinger◦ provisional | NYG | TE | 28 | 46.4% | 52.8% | 48.4% – 57.1% |
| 53 | Tyler Higbee◦ provisional | LA | TE | 70 | 50.0% | 52.7% | 48.6% – 56.8% |
| 54 | Jonnu Smith◦ provisional | ATL | TE | 70 | 50.0% | 52.7% | 48.6% – 56.8% |
| 55 | Luke Schoonmaker◦ provisional | DAL | TE | 15 | 40.0% | 52.7% | 48.3% – 57.1% |
| 56 | Juwan Johnson◦ provisional | NO | TE | 59 | 49.1% | 52.7% | 48.5% – 56.8% |
| 57 | Mo Alie-Cox◦ provisional | IND | TE | 23 | 43.5% | 52.7% | 48.3% – 57.0% |
| 58 | Jordan Akins◦ provisional | CLE | TE | 23 | 43.5% | 52.7% | 48.3% – 57.0% |
| 59 | Kylen Granson◦ provisional | IND | TE | 50 | 48.0% | 52.6% | 48.4% – 56.8% |
| 60 | Adam Trautman◦ provisional | DEN | TE | 35 | 45.7% | 52.6% | 48.3% – 56.9% |
| 61 | T.J. Hockenson◦ provisional | MIN | TE | 127 | 50.4% | 52.5% | 48.7% – 56.3% |
| 62 | John Bates◦ provisional | WAS | TE | 28 | 42.9% | 52.5% | 48.2% – 56.8% |
| 63 | Dalton Schultz◦ provisional | HOU | TE | 89 | 49.4% | 52.5% | 48.5% – 56.5% |
| 64 | Connor Heyward◦ provisional | PIT | TE | 34 | 44.1% | 52.4% | 48.2% – 56.7% |
| 65 | Andrew Ogletree◦ provisional | IND | TE | 21 | 38.1% | 52.4% | 48.0% – 56.8% |
| 66 | Gerald Everett◦ provisional | LAC | TE | 70 | 47.1% | 52.2% | 48.1% – 56.3% |
| 67 | Hayden Hurst◦ provisional | CAR | TE | 32 | 40.6% | 52.2% | 47.9% – 56.5% |
| 68 | Stone Smartt◦ provisional | LAC | TE | 21 | 33.3% | 52.1% | 47.8% – 56.5% |
| 69 | Zach Ertz◦ provisional | ARI | TE | 43 | 41.9% | 52.0% | 47.8% – 56.2% |
| 70 | Stephen Sullivan◦ provisional | CAR | TE | 24 | 33.3% | 52.0% | 47.6% – 56.3% |
| 71 | Irv Smith◦ provisional | CIN | TE | 26 | 34.6% | 51.9% | 47.6% – 56.3% |
| 72 | Tyler Conklin◦ provisional | NYJ | TE | 87 | 46.0% | 51.8% | 47.8% – 55.8% |
| 73 | Chig Okonkwo◦ provisional | TEN | TE | 77 | 44.2% | 51.6% | 47.5% – 55.6% |
| 74 | David Njoku◦ provisional | CLE | TE | 123 | 40.6% | 49.9% | 46.1% – 53.7% |
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).