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 · 2021
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
- 177
- 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 · 2021 · full board
| # | Player | Team | Pos | targets | Raw | Shrunk | 90% interval |
|---|---|---|---|---|---|---|---|
| 1 | Durham Smythe◦ provisional | MIA | TE | 41 | 82.9% | 71.1% | 65.9% – 76.0% |
| 2 | Albert Okwuegbunam◦ provisional | DEN | TE | 40 | 82.5% | 70.9% | 65.8% – 75.9% |
| 3 | George Kittle◦ provisional | SF | TE | 94 | 75.5% | 70.8% | 66.2% – 75.3% |
| 4 | C.J. Uzomah◦ provisional | CIN | TE | 63 | 77.8% | 70.8% | 65.9% – 75.5% |
| 5 | Hayden Hurst◦ provisional | ATL | TE | 31 | 83.9% | 70.6% | 65.3% – 75.7% |
| 6 | Maxx Williams◦ provisional | ARI | TE | 17 | 94.1% | 70.6% | 65.1% – 75.8% |
| 7 | Noah Fant◦ provisional | DEN | TE | 91 | 74.7% | 70.5% | 65.8% – 75.0% |
| 8 | Anthony Firkser◦ provisional | TEN | TE | 43 | 79.1% | 70.4% | 65.3% – 75.4% |
| 9 | Pat Freiermuth◦ provisional | PIT | TE | 80 | 75.0% | 70.4% | 65.6% – 75.0% |
| 10 | Gerald Everett◦ provisional | SEA | TE | 63 | 76.2% | 70.4% | 65.5% – 75.1% |
| 11 | Dalton Schultz◦ provisional | DAL | TE | 106 | 73.6% | 70.3% | 65.8% – 74.7% |
| 12 | Marcedes Lewis◦ provisional | GB | TE | 28 | 82.1% | 70.2% | 64.9% – 75.3% |
| 13 | Geoff Swaim◦ provisional | TEN | TE | 40 | 77.5% | 70.0% | 64.8% – 75.0% |
| 14 | Will Dissly◦ provisional | SEA | TE | 26 | 80.8% | 69.9% | 64.5% – 75.1% |
| 15 | Stephen Anderson◦ provisional | LAC | TE | 19 | 84.2% | 69.9% | 64.4% – 75.1% |
| 16 | John Bates◦ provisional | WAS | TE | 25 | 80.0% | 69.8% | 64.3% – 75.0% |
| 17 | Dallas Goedert◦ provisional | PHI | TE | 77 | 72.7% | 69.7% | 64.8% – 74.3% |
| 18 | Josiah Deguara◦ provisional | GB | TE | 33 | 75.8% | 69.5% | 64.2% – 74.6% |
| 19 | Tyler Higbee◦ provisional | LA | TE | 85 | 71.8% | 69.4% | 64.7% – 74.0% |
| 20 | Zach Gentry◦ provisional | PIT | TE | 25 | 76.0% | 69.3% | 63.8% – 74.5% |
| 21 | Harrison Bryant◦ provisional | CLE | TE | 28 | 75.0% | 69.2% | 63.8% – 74.4% |
| 22 | T.J. Hockenson◦ provisional | DET | TE | 86 | 70.9% | 69.2% | 64.4% – 73.8% |
| 23 | Jordan Akins◦ provisional | HOU | TE | 33 | 72.7% | 69.0% | 63.7% – 74.1% |
| 24 | Tyler Conklin◦ provisional | MIN | TE | 87 | 70.1% | 68.9% | 64.2% – 73.5% |
| 25 | Adam Shaheen◦ provisional | MIA | TE | 16 | 75.0% | 68.9% | 63.3% – 74.2% |
| 26 | MyCole Pruitt◦ provisional | TEN | TE | 19 | 73.7% | 68.8% | 63.3% – 74.2% |
| 27 | Logan Thomas◦ provisional | WAS | TE | 25 | 72.0% | 68.8% | 63.3% – 74.0% |
| 28 | Tommy Sweeney◦ provisional | BUF | TE | 12 | 75.0% | 68.7% | 63.1% – 74.2% |
| 29 | Donald Parham◦ provisional | LAC | TE | 28 | 71.4% | 68.7% | 63.3% – 74.0% |
| 30 | Brevin Jordan◦ provisional | HOU | TE | 28 | 71.4% | 68.7% | 63.3% – 74.0% |
| 31 | Drew Sample◦ provisional | CIN | TE | 15 | 73.3% | 68.7% | 63.1% – 74.1% |
| 32 | Kylen Granson◦ provisional | IND | TE | 15 | 73.3% | 68.7% | 63.1% – 74.1% |
| 33 | Dawson Knox◦ provisional | BUF | TE | 71 | 69.0% | 68.5% | 63.6% – 73.3% |
| 34 | Brock Wright◦ provisional | DET | TE | 17 | 70.6% | 68.5% | 62.9% – 73.9% |
| 35 | Travis Kelce◦ provisional | KC | TE | 134 | 68.7% | 68.5% | 64.1% – 72.7% |
| 36 | Blake Bell◦ provisional | KC | TE | 13 | 69.2% | 68.4% | 62.7% – 73.8% |
| 37 | James O'Shaughnessy◦ provisional | JAX | TE | 35 | 68.6% | 68.4% | 63.0% – 73.5% |
| 38 | Foster Moreau◦ provisional | LV | TE | 44 | 68.2% | 68.3% | 63.1% – 73.3% |
| 39 | David Njoku◦ provisional | CLE | TE | 53 | 67.9% | 68.2% | 63.1% – 73.2% |
| 40 | Pharaoh Brown◦ provisional | HOU | TE | 34 | 67.7% | 68.2% | 62.9% – 73.4% |
| 41 | Eric Saubert◦ provisional | DEN | TE | 12 | 66.7% | 68.2% | 62.5% – 73.7% |
| 42 | Eric Ebron◦ provisional | PIT | TE | 18 | 66.7% | 68.2% | 62.6% – 73.5% |
| 43 | O.J. Howard◦ provisional | TB | TE | 21 | 66.7% | 68.1% | 62.6% – 73.5% |
| 44 | Mark Andrews◦ provisional | BAL | TE | 158 | 67.7% | 68.0% | 63.8% – 72.2% |
| 45 | Blake Jarwin◦ provisional | DAL | TE | 17 | 64.7% | 68.0% | 62.4% – 73.4% |
| 46 | Hunter Henry◦ provisional | NE | TE | 75 | 66.7% | 67.8% | 62.9% – 72.6% |
| 47 | Tyler Kroft◦ provisional | NYJ | TE | 25 | 64.0% | 67.8% | 62.3% – 73.1% |
| 48 | Dan Arnold◦ provisional | JAX | TE | 53 | 66.0% | 67.8% | 62.6% – 72.8% |
| 49 | Kyle Rudolph◦ provisional | NYG | TE | 40 | 65.0% | 67.7% | 62.4% – 72.8% |
| 50 | Nick Vannett◦ provisional | NO | TE | 15 | 60.0% | 67.7% | 62.0% – 73.1% |
| 51 | Jacob Hollister◦ provisional | JAX | TE | 15 | 60.0% | 67.7% | 62.0% – 73.1% |
| 52 | Josh Oliver◦ provisional | BAL | TE | 15 | 60.0% | 67.7% | 62.0% – 73.1% |
| 53 | Ryan Griffin◦ provisional | NYJ | TE | 42 | 64.3% | 67.5% | 62.3% – 72.6% |
| 54 | Jack Doyle◦ provisional | IND | TE | 45 | 64.4% | 67.5% | 62.3% – 72.6% |
| 55 | Robert Tonyan◦ provisional | GB | TE | 29 | 62.1% | 67.4% | 62.0% – 72.7% |
| 56 | Juwan Johnson◦ provisional | NO | TE | 22 | 59.1% | 67.3% | 61.8% – 72.7% |
| 57 | Zach Ertz◦ provisional | ARI | TE | 113 | 65.5% | 67.2% | 62.6% – 71.7% |
| 58 | Jimmy Graham◦ provisional | CHI | TE | 24 | 58.3% | 67.1% | 61.6% – 72.5% |
| 59 | Ian Thomas◦ provisional | CAR | TE | 30 | 60.0% | 67.1% | 61.7% – 72.4% |
| 60 | Jonnu Smith◦ provisional | NE | TE | 45 | 62.2% | 67.1% | 61.8% – 72.2% |
| 61 | Cole Kmet◦ provisional | CHI | TE | 93 | 64.5% | 67.0% | 62.2% – 71.6% |
| 62 | Adam Trautman◦ provisional | NO | TE | 44 | 61.4% | 66.9% | 61.7% – 72.0% |
| 63 | Mike Gesicki◦ provisional | MIA | TE | 113 | 64.6% | 66.9% | 62.3% – 71.3% |
| 64 | N'Keal Harry◦ provisional | NE | TE | 22 | 54.5% | 66.8% | 61.2% – 72.2% |
| 65 | Evan Engram◦ provisional | NYG | TE | 73 | 63.0% | 66.8% | 61.8% – 71.6% |
| 66 | Austin Hooper◦ provisional | CLE | TE | 62 | 61.3% | 66.5% | 61.4% – 71.4% |
| 67 | Ricky Seals-Jones◦ provisional | WAS | TE | 51 | 58.8% | 66.2% | 61.0% – 71.3% |
| 68 | Tommy Tremble◦ provisional | CAR | TE | 36 | 55.6% | 66.2% | 60.8% – 71.4% |
| 69 | Rob Gronkowski◦ provisional | TB | TE | 89 | 61.8% | 66.1% | 61.3% – 70.8% |
| 70 | Kyle Pitts◦ provisional | ATL | TE | 110 | 61.8% | 65.8% | 61.2% – 70.4% |
| 71 | Mo Alie-Cox◦ provisional | IND | TE | 45 | 53.3% | 65.3% | 60.0% – 70.5% |
| 72 | Darren Waller◦ provisional | LV | TE | 93 | 59.1% | 65.2% | 60.3% – 69.9% |
| 73 | Cameron Brate◦ provisional | TB | TE | 57 | 52.6% | 64.5% | 59.3% – 69.6% |
| 74 | Jared Cook◦ provisional | LAC | TE | 85 | 56.5% | 64.5% | 59.6% – 69.3% |
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).