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 · 2017

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
150
targets to trust the number
shrunkrawthe shrink90% CI
40%45%50%55%60%TE avg1Hunter HenryLAC · 6255.2%2Rob GronkowskiNE · 10654.3%3Zach ErtzPHI · 11152.9%4Rhett EllisonNYG · 3252.8%5Trey BurtonPHI · 3251.7%6Jason WittenDAL · 8751.5%7Jack DoyleIND · 10951.4%8Benjamin WatsonBAL · 8051.3%9Levine ToiloloATL · 1451.3%10Adam ShaheenCHI · 1451.3%11Coby FleenerNO · 3051.1%12Travis KelceKC · 12251.1%13Cameron BrateTB · 7751.1%14Jared CookLV · 8650.9%15Jordan ReedWAS · 3550.8%16O.J. HowardTB · 3950.8%17David MorganMIN · 1250.7%18Nick O'LearyBUF · 3250.6%19Tyler KroftCIN · 6250.5%20C.J. UzomahCIN · 1550.3%21Nick VannettSEA · 1550.3%22Kyle RudolphMIN · 8150.2%23Delanie WalkerTEN · 11150.2%24Martellus BennettNE · 4450.0%25Darren FellsDET · 2650.0%26Jordan MatthewsBUF · 3650.0%27C.J. FiedorowiczHOU · 2250.0%28Nick BoyleBAL · 3850.0%29Daniel BrownCHI · 2050.0%30Jesse JamesPIT · 6349.8%31Brandon WilliamsIND · 1749.7%32Richard RodgersGB · 1949.7%33Austin TraylorDEN · 1349.7%

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 · 2017 · full board

Target success rate leaderboard for the 2017 NFL season, shrunk by empirical Bayes with 90% credible intervals, ranked by the shrunk estimate.
#PlayerTeamPostargetsRawShrunk90% interval
1Hunter Henry◦ provisionalLACTE6269.3%55.2%49.6%60.8%
2Rob Gronkowski◦ provisionalNETE10661.3%54.3%49.2%59.4%
3Zach Ertz◦ provisionalPHITE11157.7%52.9%47.8%58.0%
4Rhett Ellison◦ provisionalNYGTE3268.8%52.8%46.7%58.9%
5Trey Burton◦ provisionalPHITE3262.5%51.7%45.6%57.8%
6Jason Witten◦ provisionalDALTE8755.2%51.5%46.2%56.8%
7Jack Doyle◦ provisionalINDTE10954.1%51.4%46.3%56.5%
8Benjamin Watson◦ provisionalBALTE8055.0%51.3%45.9%56.7%
9Levine Toilolo◦ provisionalATLTE1471.4%51.3%44.9%57.7%
10Adam Shaheen◦ provisionalCHITE1471.4%51.3%44.9%57.7%
11Coby Fleener◦ provisionalNOTE3060.0%51.1%45.0%57.3%
12Travis Kelce◦ provisionalKCTE12253.3%51.1%46.1%56.1%
13Cameron Brate◦ provisionalTBTE7754.5%51.1%45.7%56.6%
14Jared Cook◦ provisionalLVTE8653.5%50.9%45.5%56.2%
15Jordan Reed◦ provisionalWASTE3557.1%50.8%44.8%56.9%
16O.J. Howard◦ provisionalTBTE3956.4%50.8%44.9%56.8%
17David Morgan◦ provisionalMINTE1266.7%50.7%44.2%57.1%
18Nick O'Leary◦ provisionalBUFTE3256.3%50.6%44.5%56.7%
19Tyler Kroft◦ provisionalCINTE6253.2%50.5%44.9%56.1%
20C.J. Uzomah◦ provisionalCINTE1560.0%50.3%44.0%56.7%
21Nick Vannett◦ provisionalSEATE1560.0%50.3%44.0%56.7%
22Kyle Rudolph◦ provisionalMINTE8151.8%50.2%44.8%55.6%
23Delanie Walker◦ provisionalTENTE11151.3%50.2%45.1%55.3%
24Martellus Bennett◦ provisionalNETE4452.3%50.0%44.1%55.9%
25Darren Fells◦ provisionalDETTE2653.8%50.0%43.9%56.2%
26Jordan Matthews◦ provisionalBUFTE3652.8%50.0%44.0%56.1%
27C.J. Fiedorowicz◦ provisionalHOUTE2254.5%50.0%43.8%56.3%
28Nick Boyle◦ provisionalBALTE3852.6%50.0%44.0%56.0%
29Daniel Brown◦ provisionalCHITE2055.0%50.0%43.7%56.3%
30Jesse James◦ provisionalPITTE6350.8%49.8%44.2%55.4%
31Austin Hooper◦ provisionalATLTE6550.8%49.8%44.2%55.4%
32Garrett Celek◦ provisionalSFTE3351.5%49.8%43.7%55.8%
33Brandon Williams◦ provisionalINDTE1752.9%49.7%43.4%56.1%
34Richard Rodgers◦ provisionalGBTE1952.6%49.7%43.4%56.1%
35Austin Traylor◦ provisionalDENTE1353.8%49.7%43.3%56.2%
36Julius Thomas◦ provisionalMIATE6250.0%49.6%43.9%55.2%
37Jermaine Gresham◦ provisionalARITE4650.0%49.5%43.7%55.4%
38Virgil Green◦ provisionalDENTE2250.0%49.5%43.2%55.7%
39Vance McDonald◦ provisionalPITTE2450.0%49.5%43.2%55.7%
40James O'Shaughnessy◦ provisionalJAXTE2450.0%49.5%43.2%55.7%
41Anthony Fasano◦ provisionalMIATE1650.0%49.4%43.1%55.8%
42Ed Dickson◦ provisionalCARTE4949.0%49.3%43.5%55.1%
43Zach Miller◦ provisionalCHITE3548.6%49.2%43.2%55.3%
44Luke Willson◦ provisionalSEATE2245.5%48.9%42.6%55.1%
45Antonio Gates◦ provisionalLACTE5347.2%48.8%43.0%54.6%
46Marcedes Lewis◦ provisionalJAXTE4946.9%48.8%43.0%54.6%
47Charles Clay◦ provisionalBUFTE7447.3%48.7%43.2%54.2%
48Troy Niklas◦ provisionalARITE2343.5%48.6%42.4%54.8%
49Clive Walford◦ provisionalLVTE1338.5%48.5%42.1%54.9%
50Demetrius Harris◦ provisionalKCTE3644.4%48.4%42.4%54.4%
51Jimmy Graham◦ provisionalSEATE9846.9%48.4%43.2%53.6%
52Brent Celek◦ provisionalPHITE2441.7%48.3%42.1%54.5%
53Maxx Williams◦ provisionalBALTE1838.9%48.3%41.9%54.6%
54Jeff Heuerman◦ provisionalDENTE1838.9%48.3%41.9%54.6%
55Ryan Griffin◦ provisionalHOUTE2740.7%48.1%41.9%54.2%
56Niles Paul◦ provisionalWASTE1936.8%48.0%41.7%54.3%
57Ross Travis◦ provisionalINDTE1735.3%47.9%41.6%54.3%
58George Kittle◦ provisionalSFTE6344.4%47.9%42.3%53.5%
59Ricky Seals-Jones◦ provisionalARITE2839.3%47.8%41.6%53.9%
60Vernon Davis◦ provisionalWASTE6943.5%47.5%42.0%53.1%
61Dion Sims◦ provisionalCHITE2937.9%47.5%41.4%53.7%
62A.J. Derby◦ provisionalMIATE4040.0%47.4%41.5%53.4%
63Darrell Daniels◦ provisionalINDTE1323.1%47.3%40.9%53.7%
64Jonnu Smith◦ provisionalTENTE3036.7%47.3%41.2%53.4%
65Tyler Higbee◦ provisionalLATE4540.0%47.2%41.4%53.1%
66Seth DeValve◦ provisionalCLETE5841.4%47.1%41.5%52.8%
67Josh Hill◦ provisionalNOTE2231.8%47.1%40.9%53.4%
68Stephen Anderson◦ provisionalHOUTE5240.4%47.1%41.3%52.8%
69Austin Seferian-Jenkins◦ provisionalNYJTE7441.9%46.9%41.4%52.4%
70Greg Olsen◦ provisionalCARTE3836.8%46.8%40.9%52.8%
71Eric Ebron◦ provisionalDETTE8641.9%46.6%41.3%52.0%
72Dwayne Allen◦ provisionalNETE2227.3%46.6%40.3%52.8%
73Lance Kendricks◦ provisionalGBTE3534.3%46.5%40.5%52.6%
74Gerald Everett◦ provisionalLATE3231.3%46.2%40.1%52.3%
75David Njoku◦ provisionalCLETE6036.7%45.7%40.1%51.4%
76Evan Engram◦ provisionalNYGTE11540.9%45.7%40.7%50.7%

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).