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.

Rush success rate · 2019

Share of carries with positive EPA. Blocking & scheme are large, un-controlled inputs.

Beats raw by
19.4%
lower out-of-sample error
RMSE raw → shrunk
9.1% → 7.3%
odd vs. even weeks
Split-half reliability
0.20
how repeatable the raw stat is
Stabilizes at
519
carrys to trust the number
shrunkrawthe shrink90% CI
34%36%38%40%42%avg1Mark IngramBAL · 20440.2%2Todd GurleyLA · 22439.6%3Aaron JonesGB · 23639.4%4Ezekiel ElliottDAL · 30139.3%5Marlon MackIND · 24839.2%6Gus EdwardsBAL · 13339.0%7Kenyan DrakeARI · 17039.0%8Chris CarsonSEA · 27839.0%9Justin JacksonLAC · 2938.9%10Derrick HenryTEN · 30338.9%11Tony PollardDAL · 8738.8%12Jordan WilkinsIND · 5138.7%13Jordan HowardPHI · 11938.6%14Alvin KamaraNO · 17138.6%15Devin SingletaryBUF · 15138.4%16Melvin GordonLAC · 16238.4%17Phillip LindsayDEN · 22538.4%18Sony MichelNE · 24738.3%19Josh JacobsLV · 24238.3%20Carlos HydeHOU · 24538.3%21Rex BurkheadNE · 6538.3%22Qadree OllisonATL · 2238.1%23LeSean McCoyKC · 10138.1%24Jamaal WilliamsGB · 10738.1%25Rashaad PennySEA · 6538.1%26Nyheim HinesIND · 5238.1%27Travis HomerSEA · 1838.1%28Derrius GuiceWAS · 4238.0%29Raheem MostertSF · 13738.0%30Dalvin CookMIN · 25038.0%31Jeff WilsonSF · 2738.0%32Spencer WareKC · 1738.0%33Cordarrelle Patte…CHI · 1738.0%34T.J. YeldonBUF · 1737.8%35Brandon BoldenNE · 1537.7%

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.

Rush success rate · 2019 · full board

Rush success rate leaderboard for the 2019 NFL season, shrunk by empirical Bayes with 90% credible intervals, ranked by the shrunk estimate.
#PlayerTeamPoscarrysRawShrunk90% interval
1Mark Ingram◦ provisionalBALRB20446.6%40.2%37.2%43.2%
2Todd Gurley◦ provisionalLARB22444.2%39.6%36.7%42.6%
3Aaron Jones◦ provisionalGBRB23643.2%39.4%36.5%42.3%
4Ezekiel Elliott◦ provisionalDALRB30142.2%39.3%36.5%42.1%
5Marlon Mack◦ provisionalINDRB24842.3%39.2%36.3%42.1%
6Gus Edwards◦ provisionalBALRB13344.4%39.0%35.9%42.2%
7Kenyan Drake◦ provisionalARIRB17042.9%39.0%35.9%42.0%
8Chris Carson◦ provisionalSEARB27841.4%39.0%36.1%41.8%
9Justin Jackson◦ provisionalLACRB2962.1%38.9%35.5%42.4%
10Derrick Henry◦ provisionalTENRB30340.9%38.9%36.1%41.7%
11Tony Pollard◦ provisionalDALRB8746.0%38.8%35.6%42.1%
12Jordan Wilkins◦ provisionalINDRB5149.0%38.7%35.3%42.0%
13Jordan Howard◦ provisionalPHIRB11942.9%38.6%35.5%41.8%
14Alvin Kamara◦ provisionalNORB17141.5%38.6%35.6%41.7%
15Devin Singletary◦ provisionalBUFRB15141.1%38.4%35.4%41.5%
16Melvin Gordon◦ provisionalLACRB16240.7%38.4%35.3%41.5%
17Phillip Lindsay◦ provisionalDENRB22540.0%38.4%35.4%41.3%
18Sony Michel◦ provisionalNERB24739.7%38.3%35.4%41.2%
19Josh Jacobs◦ provisionalLVRB24239.7%38.3%35.4%41.2%
20Carlos Hyde◦ provisionalHOURB24539.6%38.3%35.4%41.2%
21Rex Burkhead◦ provisionalNERB6543.1%38.3%35.0%41.6%
22Qadree Ollison◦ provisionalATLRB2250.0%38.1%34.7%41.6%
23LeSean McCoy◦ provisionalKCRB10140.6%38.1%34.9%41.4%
24Jamaal Williams◦ provisionalGBRB10740.2%38.1%34.9%41.3%
25Rashaad Penny◦ provisionalSEARB6541.5%38.1%34.8%41.4%
26Nyheim Hines◦ provisionalINDRB5242.3%38.1%34.8%41.4%
27Travis Homer◦ provisionalSEARB1850.0%38.1%34.6%41.5%
28Derrius Guice◦ provisionalWASRB4242.9%38.0%34.7%41.4%
29Raheem Mostert◦ provisionalSFRB13739.4%38.0%34.9%41.1%
30Dalvin Cook◦ provisionalMINRB25038.8%38.0%35.2%40.9%
31Jeff Wilson◦ provisionalSFRB2744.4%38.0%34.6%41.4%
32Christian McCaffrey◦ provisionalCARRB28838.5%38.0%35.2%40.8%
33Spencer Ware◦ provisionalKCRB1747.1%38.0%34.5%41.4%
34Cordarrelle Patterson◦ provisionalCHIRB1747.1%38.0%34.5%41.4%
35Nick Chubb◦ provisionalCLERB29838.3%37.9%35.1%40.7%
36Boston Scott◦ provisionalPHIRB6139.3%37.8%34.5%41.2%
37Saquon Barkley◦ provisionalNYGRB21738.3%37.8%34.9%40.8%
38T.J. Yeldon◦ provisionalBUFRB1741.2%37.8%34.3%41.2%
39Jonathan Williams◦ provisionalINDRB4938.8%37.8%34.4%41.1%
40Darrel Williams◦ provisionalKCRB4139.0%37.8%34.4%41.1%
41Kerryon Johnson◦ provisionalDETRB11338.0%37.7%34.6%40.9%
42Brandon Bolden◦ provisionalNERB1540.0%37.7%34.3%41.2%
43Jalen Richard◦ provisionalLVRB3938.5%37.7%34.4%41.1%
44Wayne Gallman◦ provisionalNYGRB2937.9%37.7%34.3%41.1%
45Mark Walton◦ provisionalMIARB5337.7%37.7%34.4%41.0%
46Latavius Murray◦ provisionalNORB14637.7%37.6%34.6%40.8%
47Kerrith Whyte◦ provisionalPITRB2437.5%37.6%34.3%41.1%
48Kareem Hunt◦ provisionalCLERB4337.2%37.6%34.3%41.0%
49Matt Breida◦ provisionalSFRB12337.4%37.6%34.5%40.8%
50Ito Smith◦ provisionalATLRB2236.4%37.6%34.2%41.0%
51Jon Hilliman◦ provisionalNYGRB3036.7%37.6%34.2%41.0%
52Myles Gaskin◦ provisionalMIARB3636.1%37.5%34.2%40.9%
53James Conner◦ provisionalPITRB11637.1%37.5%34.4%40.7%
54Austin Ekeler◦ provisionalLACRB13237.1%37.5%34.4%40.7%
55Elijhaa Penny◦ provisionalNYGRB1533.3%37.5%34.1%41.0%
56Darwin Thompson◦ provisionalKCRB3735.1%37.5%34.1%40.9%
57C.J. Anderson◦ provisionalDETRB1631.3%37.5%34.0%40.9%
58C.J. Prosise◦ provisionalSEARB2433.3%37.5%34.1%40.9%
59Tarik Cohen◦ provisionalCHIRB6435.9%37.5%34.2%40.8%
60Reggie Bonnafon◦ provisionalCARRB1631.3%37.5%34.0%40.9%
61Joe Mixon◦ provisionalCINRB27837.0%37.4%34.6%40.3%
62Dion Lewis◦ provisionalTENRB5435.2%37.4%34.1%40.8%
63Brian Hill◦ provisionalATLRB7835.9%37.4%34.2%40.7%
64J.D. McKissic◦ provisionalDETRB3834.2%37.4%34.1%40.8%
65Bo Scarbrough◦ provisionalDETRB8936.0%37.4%34.2%40.6%
66Miles Sanders◦ provisionalPHIRB18036.7%37.4%34.4%40.4%
67Chase Edmonds◦ provisionalARIRB6035.0%37.4%34.1%40.7%
68Ty Johnson◦ provisionalDETRB6334.9%37.4%34.1%40.7%
69Ameer Abdullah◦ provisionalMINRB2330.4%37.3%34.0%40.8%
70Darren Sproles◦ provisionalPHIRB1827.8%37.3%33.9%40.8%
71Tra Carson◦ provisionalDETRB1827.8%37.3%33.9%40.8%
72James White◦ provisionalNERB6934.8%37.3%34.1%40.6%
73Malcolm Brown◦ provisionalLARB6934.8%37.3%34.1%40.6%
74Chris Thompson◦ provisionalWASRB3732.4%37.3%34.0%40.7%
75David Johnson◦ provisionalARIRB9435.1%37.3%34.1%40.5%
76Trey Edmunds◦ provisionalPITRB2227.3%37.2%33.8%40.7%
77DeAndre Washington◦ provisionalLVRB10835.2%37.2%34.1%40.4%
78Mike Boone◦ provisionalMINRB4932.6%37.2%33.9%40.6%
79Darrell Henderson◦ provisionalLARB3930.8%37.2%33.8%40.6%
80Duke Johnson◦ provisionalHOURB8333.7%37.1%33.9%40.4%
81Bilal Powell◦ provisionalNYJRB5932.2%37.1%33.8%40.4%
82Benny Snell◦ provisionalPITRB10834.3%37.1%33.9%40.3%
83Damien Williams◦ provisionalKCRB11134.2%37.0%33.9%40.2%
84Ryquell Armstead◦ provisionalJAXRB3525.7%36.9%33.6%40.3%
85Leonard Fournette◦ provisionalJAXRB26635.3%36.9%34.1%39.7%
86Wendell Smallwood◦ provisionalWASRB2218.2%36.9%33.5%40.3%
87Justice Hill◦ provisionalBALRB5829.3%36.8%33.5%40.1%
88Alexander Mattison◦ provisionalMINRB10032.0%36.7%33.6%39.9%
89Ronald Jones◦ provisionalTBRB17233.7%36.7%33.7%39.7%
90Frank Gore◦ provisionalBUFRB16633.1%36.5%33.6%39.6%
91Royce Freeman◦ provisionalDENRB13231.8%36.5%33.4%39.6%
92Giovani Bernard◦ provisionalCINRB5324.5%36.4%33.1%39.8%
93Le'Veon Bell◦ provisionalNYJRB24533.5%36.3%33.5%39.2%
94Jaylen Samuels◦ provisionalPITRB6625.8%36.3%33.1%39.6%
95Tevin Coleman◦ provisionalSFRB13730.7%36.2%33.1%39.3%
96David Montgomery◦ provisionalCHIRB24233.1%36.2%33.4%39.1%
97Peyton Barber◦ provisionalTBRB15531.0%36.1%33.1%39.2%
98Adrian Peterson◦ provisionalWASRB21132.2%36.1%33.2%39.0%
99Kalen Ballage◦ provisionalMIARB7324.7%36.0%32.8%39.3%
100Devonta Freeman◦ provisionalATLRB18430.4%35.8%32.8%38.8%
101Patrick Laird◦ provisionalMIARB6219.4%35.7%32.5%39.0%

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).