- There are real differences: among referees with 30 or more matches, yellows per match range from 2.69 to 4.77.
- But with few matches there is a lot of noise: with 30 matches the margin of error is about ±0.36 yellows.
- The league weighs more than the referee: between leagues the average runs from 3.03 to 5.19; within a league, the referee moves it by a few tenths.
How much does the referee influence cards?
It is the question everyone looking at the cards market asks.
The short answer is: somewhat, and less than it seems.
In this article we measure it with real data.
We used the public files from football-data.co.uk.
The referee is listed in only three leagues: the Premier League, the Championship and the Scottish league.
We took the seasons from 2018/19 through 2026/27, which is in progress.
That is 8,360 matches with a referee and yellow cards recorded.
There are 163 referee-and-league combinations.
We kept those who officiated 30 matches or more in a single league: that is 86 referees and 7,758 matches.
Honest limit: these files do not include card odds.
That is why we cannot say which market pays.
We can say how much yellows change depending on who officiates.
This is not betting advice and is for adults 18 and over only.
How different are referees from one another?
Very different at the extremes.
The table shows, for each league, the referee with the fewest and the most yellows per match.
Only referees with 30 matches or more are included.
| League | Referees | League average | Lowest | Highest |
|---|---|---|---|---|
| Premier League | 26 | 3.66 | 2.69 | 4.41 |
| Championship | 41 | 3.64 | 2.96 | 4.77 |
| Scottish league | 19 | 3.68 | 2.81 | 4.39 |
In the Premier League, the most lenient referee averages 2.69 yellows.
The strictest, 4.41.
That is a difference of 1.7 yellows per match.
On average, a referee sits 0.34 yellows away from their league's average.
The spread of all referees relative to their league is 0.42 yellows.
But that spread is not all real, as we see next.
How much of that difference is noise?
This is the part almost nobody talks about.
Yellows per match swing a lot from one match to the next.
In these three leagues the typical deviation is 1.97 yellows.
| Referee's matches | Standard error | Reading |
|---|---|---|
| 30 matches | ±0.36 | Very noisy |
| 80 matches | ±0.22 | Acceptable |
With 30 matches, a referee whose average looks 0.36 higher than the league can be perfectly normal.
That is the standard error, calculated as the deviation divided by the square root of the number of matches.
With 80 matches the margin drops to ±0.22.
We ran an extra test.
We shuffled yellow cards at random among matches in the same league, 300 times, and measured the spread that appears with no real difference between referees.
It averaged 0.23, with a maximum of 0.29.
The real spread was 0.42.
That is larger than anything pure chance produced.
Verdict: partly true, but real.
Referees do differ, although noise inflates the differences.
Is a strict referee always strict?
Another frequent doubt is whether the referee data lasts.
We compared 2023/24 and 2024/25 against 2025/26 and 2026/27.
That included 44 referees with at least 15 matches in each period.
The correlation between the two halves was 0.29.
It is positive, but low.
In other words, some of a referee's personality holds, but a lot changes from one period to the next.
Verdict: partly true.
It works as a hint, not as a rule.
The referee's style matters less than the match context: who is playing, what is at stake and what the atmosphere is like.
What weighs more, the referee or the league?
The league, by far.
In our full sample of 16 leagues, the average of yellows ranges from 3.03 in the Netherlands to 5.19 in Portugal.
That is a difference of 2.2 yellows.
Within a league, the referee moves it by a few tenths.
Only at the extremes does it approach a full yellow.
That is why looking at the league first saves you from many mistakes.
For the detail on each league, see the table in card stats from 39,856 matches.
Myth 1: "with a few matches you already know what a referee is like"
False.
With fewer than 30 matches the average is very unstable.
That is why we set a floor of 30 matches, and even then the standard error is ±0.36.
If an article or an AI tells you a referee "hands out twice as many cards," ask how many matches they have officiated.
If it is ten, it is an anecdote.
Myth 2: "the referee favors the home team"
It cannot be known with this data.
The files show that the away team gets more yellows than the home team.
But we cannot separate the referee's bias from each team's playing style.
What about fouls? Does the referee whistle them differently?
The files also include fouls committed.
If we look at each referee's fouls per match, the difference from their league ranges from −3.3 to +3.2.
In the Premier League, from 18.1 to 23.4 fouls per match, against a league average of 21.4.
A lenient referee whistles fewer fouls, and that tends to lower yellows.
A strict referee does the opposite.
The link is not perfect: the correlation between fouls and yellows per match is 0.41.
Fouls work as a complementary signal.
You will find more detail on what they can and cannot tell you in referees, penalties and VAR.
How do you use the referee data, step by step?
The "example" in step 4 comes from a simple calculation.
It starts from the 0.42 total spread and the 0.23 of pure-chance noise, and gives a weight of roughly 0.7 for a referee with 79 matches.
It is an estimate, not a law.
How do I ask an AI for a referee's history?
An AI with web search can gather a referee's history faster than you can.
But you have to ask for the data with a clear source and clear limits.
Use this prompt as a base.
Act as a football data analyst. I will give you a referee's name and a league: [REFEREE] in [LEAGUE]. Search the web for their record over the last three seasons. Give me: matches officiated, yellow cards per match, red cards per match and fouls per match. For each figure, state the source with a link and the date you accessed it. Compare with the league average and tell me how many matches are in the sample. If you cannot find a figure, write "no data" instead of estimating it. Do not make any betting recommendation.
Then verify one or two figures by hand.
If the AI cannot give you the source, do not use the number.
We tell you more about how to compare tools in the best AI for betting and about building your own assistant in create your own AI.
What to read next
If you want more on referees, continue with referees, penalties and VAR in betting.
It is the second part of this article and explains what data is missing.
To see cards league by league, there is card stats: myths and data.
And to learn the market from scratch, see the card betting guide.
The full picture of AI applied to betting is in the main guide.
To see how we measure our own picks, there is the Betting Journal.
Betting carries risk and is for adults 18 and over only.
If you feel you are losing control, ask for help.
Past data guarantees nothing going forward.
Frequently asked questions
Does the referee influence the number of cards?
Yes, but less than people think.
Across 86 English and Scottish referees, the average ranges from 2.69 to 4.77 yellows per match.
Part of that gap is chance.
Use it as one signal among several.
How many matches do I need to trust a referee's numbers?
More than 30, and even then with care.
With 30 matches the typical error is about ±0.36 yellows.
With 80 it drops to about ±0.22.
The fewer the matches, the more you should distrust the average.
Is a strict referee still strict the following season?
In part.
We compared two periods with 44 referees and the correlation was 0.29.
It is positive but low: something stays stable and a lot changes.
What weighs more, the referee or the league?
The league.
Across the 16 leagues in our sample the average ranges from 3.03 to 5.19 yellows.
Within a league, the referee moves it by a few tenths, although at the extremes it reaches more than one yellow.
Do you know the referee before the match?
Usually yes, a few days ahead.
The official list depends on each league.
Always confirm the appointment before using the data.
Can I use this data to tell if card betting is profitable?
No.
The files we used do not include card odds.
We measured how yellows are distributed, not how much each market pays.