Terrible-2.60 contribution32nd of 34 picks for this GM
Games: direct team-season record. Starts: direct team-season record.
Long-career hit
Leonard Williams became a 11-season, 175-game NFL player after going #6. The retained-value model still ranks him 32nd of 34 for Mike Maccagnan because NYJ only got 71 of those games.
His retained-value contribution score (opportunity plus quality plus recognition, with related outcomes collapsed inside each bucket) is -2.60. This is a per-pick accountability metric, separate from Mike Maccagnan's leaderboard composite θ.
Full-career context prevents this from being labeled as a bad talent pick: the negative model score is about retained team value, not whether the draft identified an NFL-caliber player.
The walkthrough below shows which bucket carried the pick: playing time, value quality, or league honors.
Opportunity
Did he earn the field?
Games played
71
on drafting team · 175 full-career
Games started
70
on drafting team · 166 full-career
Snap-z
7.3
5 modeled seasons
Quality
Was that time valuable?
On-team AV
37
modeled; full weighted AV 78
AV / 1k snaps
8.8
4,193 snaps in rate base
AV-rate z
+0.2
position-year adjusted quality
Recognition
Did the league notice?
Pro Bowls
1
3 full-career
AP1 All-Pro
0 AP1
all on drafting team · AP2 context 0
All-Decade
0
context only
Did this pick move Mike Maccagnan's grade?
The first chart removes the time axis: every Mike Maccagnan pick is sorted by retained contribution across opportunity, quality, and recognition. Leonard Williams totals -2.60. The second chart breaks that same contribution into the seasons when the evidence arrived. Opportunity and quality can move separately, which keeps high-snap, low-AV profiles from looking cleaner than they are.
Aggregate contribution by player
Incremental contribution by season
Above expectationBelow expectationLeonard WilliamsGM total -19.4Leonard Williams by season -5.5
How much retained production did Leonard Williams return vs. his slot?
Each mark below is one player's model value above slot. Light gray marks are slot-similar peers: DEs drafted within ±5 picks of Leonard Williams since 2010 (17 players).The score is the retained production/value bucket from the pick-residual model: on-team AV plus AV-rate quality, adjusted for draft slot, position, and era. Opportunity snaps and league honors are excluded.
Leonard WilliamsSlot-similar peers (17)
Cumulative value over time
Year-by-year running totals since draft. Left axis = games started. Right axis = honors (Pro Bowl / AP1 / AP2). Pre-2013 picks have no games-started line because that comes from per-game snap counts (not available for older players). Honors extend back to 1980.
Games started (cumulative)Pro BowlsAP1 All-ProAP2 All-Pro
Full-career season history (11 scored seasons)
Year
Team
GP
GS
Snaps
Snap-z
AV
AV/snap
PB
AP1
AP2
2015
NYJ
16
14
885
1.4
9
10.2
2016
NYJ
16
16
978
1.9
10
10.2
✓
2017
NYJ
16
15
954
1.9
7
7.3
2018
NYJ
16
15
957
1.8
8
8.4
2019
NYG, NYJ
15
15
825
0.5
6
7.3
2020
NYG
16
16
903
1.9
8
8.9
2021
NYG
17
16
984
2.1
9
9.1
2022
NYG
12
11
650
0.9
6
9.2
2023
NYG, SEA
18
18
986
0.9
7
7.1
2024
SEA
16
15
804
1.6
13
16.2
2025
SEA
17
17
866
1.7
17
19.6
How to read this
The current model groups outcomes into three buckets: opportunity (games, starts, snap-z), quality (on-team AV, AV/snap quality when available), and recognition (Pro Bowl, AP1).
Season rows show full-career context across every team. The draft-credit cards above contain the team-specific totals used for GM accountability. Snap-z is position/year-normalized playing time; AV/snap is shown as AV per 1,000 snaps where snap evidence exists.
AP2, full-career AV, HOF, and All-Decade are displayed as context only. They are not active outcomes in the current θ model.
Pre-2013 playing-time rows may show "—" where weekly snap evidence is unavailable; AV/snap quality is snap-era only.
