Was the referee against us?
Your team's card count says less than you think. Here is the arithmetic.
The arithmetic, so you can check it
A count of about two, arriving at random, varies by roughly its own square root. That is 1.436. What we measure is 1.418 — across 2,280 club-matches from 2023-24 onward. Simulating the same matches with the referee removed entirely gives 1.437 [1.369, 1.492] over 200 runs.
How often a gap that big happens, against how often chance makes it
| Gap in cards | Observed | Chance | Difference |
|---|---|---|---|
| 1.5+ | 30.1% | 29.4% | +0.7 |
| 2.0+ | 15.7% | 15.5% | +0.2 |
| 2.5+ | 6.0% | 6.6% | -0.6 |
| 3.0+ | 3.0% | 3.3% | -0.3 |
| 3.5+ | 1.4% | 1.9% | -0.5 |
| 4.0+ | 0.7% | 1.0% | -0.3 |
Observed is at or below chance at every threshold. There is no size of gap at which "we were booked far more than usual" happens more often than randomness alone would make it.
The biggest gaps we hold
| Club | Season | Booked | Their average | Gap | Other matches |
|---|---|---|---|---|---|
| Nottingham Forest | 2024-25 | 8 | 2.2 | +5.8 | 36 |
| Chelsea | 2024-25 | 8 | 2.5 | +5.5 | 33 |
| Crystal Palace | 2023-24 | 7 | 1.8 | +5.2 | 36 |
| Aston Villa | 2024-25 | 7 | 1.8 | +5.2 | 36 |
| Nottingham Forest | 2023-24 | 7 | 2.1 | +4.9 | 34 |
| Arsenal | 2025-26 | 6 | 1.2 | +4.8 | 34 |
| Sheffield Utd | 2023-24 | 7 | 2.5 | +4.5 | 36 |
| Arsenal | 2023-24 | 6 | 1.5 | +4.5 | 35 |
Each average is that club's own rate under every OTHER referee in the same season, so the referee being asked about is never in his own baseline. Each club-match compares the cards that club took to its own average under every OTHER referee in the same season, with at least 10 such matches.
This is not a defence of referees
It is a statement about what a card count can detect, and nothing more. A referee could be biased in ways this measure cannot see — in fouls not given, in advantage played, in which challenges draw a word instead of a card — and every number on this page would look exactly the same. What the page rules out is the specific inference people draw from the specific number they can see. It does not rule out the thing they suspect.
One measure does come out above chance
Reported because leaving it out would be the selective quoting this page argues against. Take the total cards in a match against what both clubs average, and the spread is 2.189 where chance gives 2.029 — about 0.82 cards of real excess. Some matches genuinely are carded more than the two teams' habits predict.
It still cannot carry a claim about one match. At a gap of 4+ cards it fires on about 25 matches a season, and roughly 17 of those would fire by chance anyway. Two firings in three are noise, and nothing tells you which is which. A per-match line built on it would be wrong about most of the matches it appeared on.