The NBA grades its own clutch officiating
When a game is within three points at the two-minute mark, the league reviews every officiated event in that window and publishes a verdict on each: a correct call, a correct non-call, an incorrect call, or an incorrect non-call. This is 4,608 of those reports, 2014-15–2025-26, as 89,052 graded decisions. Browse every reviewed game →
Roughly one call in eleven is graded wrong — about 1.33 per reviewed game. That is the league's own assessment, not an outside one, which cuts both ways: it is unusually candid for a sports body, and it is also the referee's employer marking the referee's homework.
How the verdicts break down
Every reviewed event falls into one of four categories. Non-calls dominate, because most of what happens in a crowded final two minutes is contact the officials decide to let go.
Show the data
| Verdict | Calls |
|---|---|
| CC | 21,827 |
| CNC | 61,113 |
| IC | 830 |
| INC | 5,282 |
Source: NBA Last Two Minute reports
What gets reviewed
Fouls are most of it. Labels are shortened — every one but the last three reads “Foul: …” in the source.
Show the data
| Call type | Calls |
|---|---|
| Personal | 27,298 |
| Shooting | 24,313 |
| Offensive | 16,779 |
| Loose Ball | 8,240 |
| Turnover: Traveling | 1,888 |
| Stoppage: Out-of-Bounds | 1,327 |
| Defense 3 Second | 1,061 |
| Instant Replay: Support Ruling | 882 |
Source: NBA Last Two Minute reports
Error rate by season
Share of graded calls the league marked incorrect.
Show the data
| Season | Incorrect calls (%) |
|---|---|
| 2015 | 13% |
| 2016 | 11.5% |
| 2017 | 8.6% |
| 2018 | 6.1% |
| 2019 | 6.6% |
| 2020 | 6.7% |
| 2021 | 6.1% |
| 2022 | 7.6% |
| 2023 | 6.1% |
| 2024 | 6.1% |
| 2025 | 4.5% |
| 2026 | 5.1% |
The 2015–2018 seasons come from the archived PDFs; later seasons from the league's JSON feed. The two were parsed by different pipelines, so small season-to-season steps are not worth reading closely.
Source: NBA Last Two Minute reports
What this can't tell you
Only close games are reviewed, so this is not a sample of NBA officiating — it is a sample of officiating under maximum pressure, in the situations most likely to be scrutinised. A blowout's final two minutes never appear here.
It also can't answer the obvious next question, which is who. The archived reports name the three officials working each game, but the league's current feed names nobody, and the endpoint where they could otherwise be looked up blocks automated clients. Per-referee analysis is therefore limited to 12 seasons (2014-15–2025-26) — old enough that most of those crews have changed.
And a verdict is a judgement, not a measurement. The reports are produced by the same organisation that employs the officials, reviewing plays in slow motion that were called in real time by someone who got one look.
Independent, non-commercial project for analysis and education — not affiliated with or endorsed by the NBA. Data from public NBA releases (pr.nba.com), tabulated by Ernst & Young. Source ↗