Bias Trends
Who keeps getting bad calls — over time?
Per-month trend of AI-judged incorrect calls for the top 8 players and teams. Rising lines mean a pattern of bias is accumulating.
2 buckets
Whistle Consistency
How closely does each official match the AI's "Truth"?
100 means every judged call matched the AI verdict. The blue → violet gradient highlights where calls deviate most.
Executive Summary · AI-generated
Flagged game: New York Knicks @ Cavaliers
50% of 4 reviewed calls in this game (2026-05-24) were judged incorrect.
Pressure Rating
Does crew accuracy hold up on the road?
Pressure = home accuracy − away accuracy. Positive values mean the crew's "Truth Accuracy" drops in hostile environments.
- Ed MalloyH 80% · A 50%+30Pressure
- Ken MauerH 100% · A 75%+25Pressure
- Jacyn GobleH 75% · A 50%+25Pressure
- Kane FitzgeraldH 100% · A 80%+20Pressure
- Zach ZarbaH 81.8% · A 68.8%+13.1Pressure
- James CapersH 87.5% · A 78.6%+8.9Pressure
Players
Top 8 players by total incorrect calls
Wrong calls per game
#
Player
GP
Calls
Wrong
Rate
Wrong / game
#1
Jeremy Sochan · SAS
61 GP · 62 calls · 22 wrong
0.36
wrong/g
1
Jeremy Sochan · SAS
61
62
22
46%
0.36
#2
Evan Mobley · CLE
67 GP · 69 calls · 20 wrong
0.30
wrong/g
2
Evan Mobley · CLE
67
69
20
34%
0.30
#3
Devin Vassell · SAS
49 GP · 50 calls · 18 wrong
0.37
wrong/g
3
Devin Vassell · SAS
49
50
18
49%
0.37
#4
Chet Holmgren · OKC
65 GP · 68 calls · 17 wrong
0.26
wrong/g
4
Chet Holmgren · OKC
65
68
17
27%
0.26
#5
Shai Gilgeous-Alexander · OKC
55 GP · 56 calls · 17 wrong
0.31
wrong/g
5
Shai Gilgeous-Alexander · OKC
55
56
17
35%
0.31
#6
Donovan Mitchell · CLE
58 GP · 60 calls · 16 wrong
0.28
wrong/g
6
Donovan Mitchell · CLE
58
60
16
31%
0.28
#7
Darius Garland · CLE
40 GP · 40 calls · 16 wrong
0.40
wrong/g
7
Darius Garland · CLE
40
40
16
50%
0.40
#8
Julius Randle · NYK
38 GP · 40 calls · 14 wrong
0.37
wrong/g
8
Julius Randle · NYK
38
40
14
37%
0.37
Teams
Top 8 teams by total incorrect calls against
Wrong calls per game
#
Team
GP
Calls
Wrong
Rate
Wrong / game
#1
CLE
121 GP · 241 calls · 62 wrong
0.51
wrong/g
1
CLE
121
241
62
32%
0.51
#2
SAS
119 GP · 228 calls · 57 wrong
0.48
wrong/g
2
SAS
119
228
57
31%
0.48
#3
OKC
111 GP · 207 calls · 55 wrong
0.50
wrong/g
3
OKC
111
207
55
32%
0.50
#4
NEW
77 GP · 144 calls · 41 wrong
0.53
wrong/g
4
NEW
77
144
41
31%
0.53
#5
DET
37 GP · 76 calls · 10 wrong
0.27
wrong/g
5
DET
37
76
10
16%
0.27
#6
MIN
24 GP · 58 calls · 10 wrong
0.42
wrong/g
6
MIN
24
58
10
18%
0.42
#7
NYK
9 GP · 14 calls · 4 wrong
0.44
wrong/g
7
NYK
9
14
4
40%
0.44
#8
LAL
9 GP · 15 calls · 3 wrong
0.33
wrong/g
8
LAL
9
15
3
20%
0.33
Wrong calls are the AI's incorrect verdicts on judged calls. Inconclusive judgements are excluded. Only entities with ≥ 5 total calls are charted.