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.

Dropback success rate · 2023

Share of dropbacks (incl. sacks & scrambles) with positive EPA.

Beats raw by
15.6%
lower out-of-sample error
RMSE raw → shrunk
5.5% → 4.6%
odd vs. even weeks
Split-half reliability
0.47
how repeatable the raw stat is
Stabilizes at
147
dropbacks to trust the number
shrunkrawthe shrink90% CI
35%40%45%50%55%avg1Brock PurdySF · 46951.5%2Tua TagovailoaMIA · 58948.6%3Josh AllenBUF · 60548.3%4Dak PrescottDAL · 63148.3%5Jared GoffDET · 63648.2%6Lamar JacksonBAL · 49447.0%7Geno SmithSEA · 52846.7%8Joe BurrowCIN · 38845.9%9Jordan LoveGB · 60945.8%10Carson WentzLA · 2745.7%11Jake BrowningCIN · 26845.7%12Jimmy GaroppoloLV · 18345.6%13Patrick MahomesKC · 62245.6%14C.J. StroudHOU · 53545.4%15Matthew StaffordLA · 55145.3%16Trevor LawrenceJAX · 60245.3%17Tyson BagentCHI · 14945.1%18Desmond RidderATL · 42045.1%19Derek CarrNO · 58145.0%20Jalen HurtsPHI · 57544.8%21Marcus MariotaPHI · 2644.8%22Kirk CousinsMIN · 32644.7%23Mason RudolphPIT · 8044.7%24Sam DarnoldSF · 5144.7%25Justin HerbertLAC · 48644.5%26Baker MayfieldTB · 61044.2%27Cooper RushDAL · 2543.9%28Jarrett StidhamDEN · 7343.9%29C.J. BeathardJAX · 6043.7%30Nick MullensMIN · 15943.6%31Brian HoyerLV · 4343.4%

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.

Dropback success rate · 2023 · full board

Dropback success rate leaderboard for the 2023 NFL season, shrunk by empirical Bayes with 90% credible intervals, ranked by the shrunk estimate.
#PlayerTeamPosdropbacksRawShrunk90% interval
1Brock PurdySFQB46954.2%51.5%48.2%54.9%
2Tua TagovailoaMIAQB58949.9%48.6%45.6%51.6%
3Josh AllenBUFQB60549.6%48.3%45.3%51.3%
4Dak PrescottDALQB63149.5%48.3%45.3%51.2%
5Jared GoffDETQB63649.4%48.2%45.3%51.1%
6Lamar JacksonBALQB49448.2%47.0%43.8%50.3%
7Geno SmithSEAQB52847.7%46.7%43.6%49.9%
8Joe BurrowCINQB38846.9%45.9%42.4%49.4%
9Jordan LoveGBQB60946.5%45.8%42.9%48.8%
10Carson Wentz◦ provisionalLAQB2759.3%45.7%39.5%51.9%
11Jake BrowningCINQB26847.0%45.7%41.7%49.7%
12Jimmy GaroppoloLVQB18347.5%45.6%41.1%50.1%
13Patrick MahomesKCQB62246.1%45.6%42.6%48.5%
14C.J. StroudHOUQB53546.0%45.4%42.3%48.5%
15Matthew StaffordLAQB55145.9%45.3%42.3%48.4%
16Trevor LawrenceJAXQB60245.9%45.3%42.3%48.3%
17Tyson BagentCHIQB14947.0%45.1%40.4%49.9%
18Desmond RidderATLQB42045.7%45.1%41.6%48.5%
19Derek CarrNOQB58145.4%45.0%42.0%48.0%
20Jalen HurtsPHIQB57545.2%44.8%41.8%47.9%
21Marcus Mariota◦ provisionalPHIQB2653.8%44.8%38.6%51.0%
22Kirk CousinsMINQB32645.4%44.7%41.0%48.5%
23Mason Rudolph◦ provisionalPITQB8047.5%44.7%39.3%50.2%
24Sam Darnold◦ provisionalSFQB5149.0%44.7%38.9%50.5%
25Justin HerbertLACQB48644.9%44.5%41.2%47.7%
26Baker MayfieldTBQB61044.4%44.2%41.2%47.2%
27Cooper Rush◦ provisionalDALQB2548.0%43.9%37.7%50.1%
28Jarrett Stidham◦ provisionalDENQB7345.2%43.9%38.4%49.4%
29C.J. Beathard◦ provisionalJAXQB6045.0%43.7%38.1%49.4%
30Nick MullensMINQB15944.0%43.6%39.0%48.3%
31Tyrod TaylorNYGQB19243.8%43.5%39.1%48.0%
32Mitchell Trubisky◦ provisionalPITQB11443.9%43.5%38.5%48.5%
33Brian Hoyer◦ provisionalLVQB4344.2%43.4%37.6%49.4%
34Joe FlaccoCLEQB21342.7%42.9%38.7%47.2%
35Tyler Huntley◦ provisionalBALQB4141.5%42.8%36.9%48.8%
36Andy Dalton◦ provisionalCARQB6141.0%42.5%37.0%48.2%
37Gardner MinshewINDQB52742.3%42.5%39.4%45.6%
38Mac JonesNEQB36842.1%42.4%38.9%46.0%
39Taylor Heinicke◦ provisionalATLQB14241.5%42.4%37.6%47.2%
40Anthony Richardson◦ provisionalINDQB9340.9%42.3%37.1%47.6%
41Russell WilsonDENQB49142.0%42.2%39.0%45.5%
42Aidan O'ConnellLVQB36741.7%42.1%38.6%45.7%
43Case Keenum◦ provisionalHOUQB5937.3%41.5%35.9%47.2%
44Kyler MurrayARIQB29040.3%41.3%37.5%45.2%
45Sam HowellWASQB67940.8%41.2%38.4%44.0%
46Jeff Driskel◦ provisionalCLEQB2931.0%41.2%35.2%47.3%
47Deshaun WatsonCLEQB18739.6%41.2%36.8%45.6%
48Easton StickLACQB18739.6%41.2%36.8%45.6%
49Will LevisTENQB28240.1%41.1%37.3%45.1%
50Davis Mills◦ provisionalHOUQB4233.3%41.0%35.2%46.9%
51Ryan TannehillTENQB26039.6%40.9%36.9%44.9%
52Kenny PickettPITQB34739.8%40.8%37.2%44.5%
53Blaine Gabbert◦ provisionalKCQB3630.6%40.7%34.8%46.7%
54Joshua DobbsMINQB45039.3%40.3%37.0%43.6%
55Jameis Winston◦ provisionalNOQB4930.6%40.1%34.4%45.9%
56Dorian Thompson-Robinson◦ provisionalCLEQB11735.0%39.6%34.7%44.6%
57Daniel JonesNYGQB19136.6%39.5%35.2%43.9%
58Drew Lock◦ provisionalSEAQB8332.5%39.4%34.1%44.7%
59Justin FieldsCHIQB41637.7%39.2%35.8%42.6%
60Brett Rypien◦ provisionalLAQB4022.5%38.8%33.0%44.7%
61Zach WilsonNYJQB41337.0%38.7%35.3%42.1%
62Clayton Tune◦ provisionalARIQB2814.3%38.6%32.6%44.7%
63PJ Walker◦ provisionalCLEQB12132.2%38.3%33.4%43.2%
64Trevor SiemianNYJQB16133.5%38.1%33.7%42.7%
65Tim Boyle◦ provisionalNYJQB8626.7%37.1%32.0%42.4%
66Bryce YoungCARQB58935.3%36.9%34.0%39.8%
67Tommy DeVitoNYGQB21532.6%36.9%32.8%41.1%
68Bailey ZappeNEQB23632.6%36.7%32.7%40.8%

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).