Added time · the second-half clock
How long do referees really add on?
Second-half added time, per referee — the on-field referee's own call. How "expected" is calculated →
Tight games get the most added time by score margin at 90'
Across 658 matches, classified by the score at the end of 90 minutes. The model already captures this — goal-margin is one of its predictors.
Added time, two ways by referee, or by team's game state at 90'
Ordered by minutes added — a way to navigate the list, not a ranking.
No referee here is detectably different from what his own matches predict: every gap sits inside
the error. Read that as can't detect, not doesn't exist — the model explains only about a
third of why added time varies, so a referee could add up to half a minute more than expected in
every match and still not stand out from it.
Why we don't crown one →
| # | Referee | Matches | Avg 2nd-half added | Raw avg | Expected | vs expected |
|---|
How to read it. Raw avg is the real second-half added time he averaged — the trustworthy number. vs expected is raw minus what the model predicts for his matches: positive means he adds a touch more than the visible stoppages suggest, negative a touch less. It's a tendency, not a verdict — most of added time (injuries, VAR-check length) is invisible to the model. 2024-25 and 2025-26 only; 2023-24 has no added-time figure.