How a club can commit the most fouls, but get the fewest yellow cards

By Tony Attwood


How do farcical transfer rumours start? – the video

First a health warning.  If you believe that one can prove anything with statistics, and that the money spent by clubs buying in data from Opta is a pathetic waste of cash which could be spent on buying more players, then this is not an article for you.

But if you are looking for an explanation of what is going on in the Premier League, this might be of interest.  Because below is the league table which has never been published before.  It shows the PL as it currently stands with Liverpool at the top and Arsenal 10th.  But instead of points and goals etc, this shows where each club stands in three other league tables.

  • First the club’s position in the league table of tackles.
  • Second the position in the league table of fouls.
  • Third the position in the league table of yellow cards.
  • Then fourth the ratio between tackles and fouls, meaning how many tackles the club puts in, in order to get a foul.   And finally, the ratio between fouls and yellows.   Meaning how many fouls the club commits in order to get a yellow card.

I have also highlighted the top three (in red) and bottom three (in brown) in each column.   Here are the highlights.  If you would like further data, such as the number of tackles and the number of fouls etc they can be found in our earlier analysis.

  • Tackles: Leicester, Southampton and WHU commit the most tackles
  • Fouls: Southampton, Everton and Watford commit the most fouls
  • Yellow cards: Arsenal, Tottenham and Watford get the most yellow cards.
  • Most tackles to number of fouls: Leicester, Liverpool and WHU commit the most tackles per foul.
  • Most fouls committed to get a yellow card: Leicester, Liverpool and Southampton.
Pos Team Tackles pos Fouls pos Yellow pos Tackle/ Fouls Fouls/ Yellows
1 Liverpool 16 20 20 1.90 8.33
2 Manchester City 18 13 7 1.43 5.21
3 Leicester City 1 16 19 2.08 9.48
4 Chelsea 7 15 11 1.82 5.31
5 Tottenham Hotspur 9 14 2 1.79 4.46
6 Sheffield United 10 7 4 1.61 5.36
7 Manchester United 15 6 7 1.45 5.68
8 Wolverhampton W 11 9 16 1.63 6.82
9 Everton 4 2 12 1.60 6.55
10 Arsenal 13 10 1 1.61 4.32
11 Burnley 19 5 12 1.29 6.14
12 Newcastle United 14 19 15 1.81 5.64
13 Southampton 2 1 16 1.67 7.73
14 Crystal Palace 8 8 14 1.67 6.50
15 Brighton and Hove A 6 12 18 1.76 6.92
16 AFC Bournemouth 20 18 5 1.55 4.62
17 Aston Villa 17 4 9 1.39 5.89
18 West Ham United 3 11 9 1.88 5.61
19 Watford 5 3 2 1.55 5.30
20 Norwich City 12 17 5 1.54 4.83

What I am looking for here are patterns – patterns which give a hint of what is going on in games.

In column one, there is no pattern.  We measure the number of tackles, and the top three tackling sides are spread out through the league table, and so are the teams that commit the least number of tackles.  The number of tackles thus give us no indication of how well a team is doing.

Leicester tackle most and are near the top of the league.  WHU are third in the tackling list but are looking at relegation.  The same variation appears with clubs that tackle the least – Man City are 18th in the tackling chart, but are second in the league.  Bournemouth tackle the least, but are near the foot.

There is also a mix in terms of fouls and yellow cards – one cannot say that fouling teams and teams with yellow cards are generally near the top or bottom of the league.

So was all that a waste of time?  Well, actually no – for the very reason that the numbers seem to be jumbled up.  What we need to do is look at the last two columns.

The penultimate column shows us that Burnley commit a foul every 1.29 tackles.  Roughly speaking if they tackle 13 times then they commit 10 fouls.  Leicester, on the other hand, would need to undertake 21 tackles to give away 10 fouls.   That is an astonishing difference.  Every Burnley tackle is 61% more likely to be a given as a foul than every Leicester tackle!

What makes this even odder is that there is no pattern discernible.  It is not as if teams that tackle more or less generally commit more or fewer fouls.

When we look at the number of fouls that lead to yellow cards, the situation is even more extreme.  Arsenal are more than twice as likely to be penalised each time they tackle, than Leicester.

Now when we have touched on this before I have seen people write in with the comment that it is because we have Xhaka, who doesn’t know how to tackle.  But Xhaka has had only five yellows this season which means he is not even in the top 20 list of players for yellow cards.  Three players in the PL have already knocked up nine cards each.  So even allowing for the matches Xhaka has missed, the fact of him getting yellows all the time is a media tale, not reality.

So how do we make sense of this?

The most obvious lesson to learn is that it is worth practising tackling to make sure tackles are not seen as fouls.  And yet… Manchester City is a club that has been winning the league of late and it seems that both they and Liverpool have learned to play with a low number of tackles.  But what is harming Man City is that the tackles that they do deliver are much more likely to be called as fouls by referees than Liverpool’s tackles.  43% more likely in fact.   Worse, when a tackle by Man City is undertaken, it is almost twice as likely to be deemed a yellow card offence as a tackle from a Leicester player.


It is self-evident that there are many different ways of setting teams up to play professional football.  It can be as a team that tackles a lot, or little, a team that fouls a lot, or rarely… there is no one set way.

But there is a huge difference in the way that referees respond to teams.  When we did the 160 game survey we found that referees were giving decisions to different teams in different ways – what one team could get away with, another would not.  And that occurred with different referees.

If you want to see this magnified, my recommendation would be to watch a Premier League game one day, and a League Two or a National League game the next day.  Theoretically, the rules are exactly the same, but in my experience what referees allow players to get away with, and not even call a foul, in League Two or the National League will often be a yellow in the Premier League.

Now it shouldn’t be like that but it is.

What these figures suggest to me is that there is, for reasons I cannot ascertain, a similar division going on within the Premier League.  What is a yellow offence when an Arsenal player does it, is not a yellow offence when a Leicester player does it.

So, according to these official figures above, Leicester commit more tackles than any other team, but commit fewer fouls than most, and only one team (Liverpool) gets fewer yellows than they do.

Now I find this extraordinary, first because the difference in the numbers is so huge, and second because clubs have got whole departments studying videos of matches.  Everyone must know that Leicester are getting fewer yellow cards while committing a huge number of tackles because their tackles are not being called fouls.

And what is even more extraordinary is that although these figures are freely available, the mass media have never once touched upon them.

The simple questions thus are: why are clubs not copying Leicester’s style and approach, and why is this issue not a subject for debate?  You may well argue that it is all happening because Leicester have adopted this policy, and have very very clever players who are trained to tackle without fouling.  Maybe.  But given that every action on the pitch is being filmed from every angle, is it really likely that Leicester, alone among the clubs, have managed to achieve this?  And if they have, surely that is an even greater reason for this to be a key topic in every football programme on TV.

On the other hand the giving of fouls for tackles is an issue that is very personal to the referee, and surely many of us will remember the way certain clubs like Bolton would come to Arsenal and just try and kick us into the ground, seemingly with impunity.

That observation is, unlike everything before, a personal one not based on data, but it is useful since it suggests that the explanation of the data is a simple one.   Referees react to different teams in different ways.  (The most obvious example of this was Bolton’s time wasting tactics which we observed for year after year).

Of course if you can find another explanation for this data, that’s fine, but for the moment different referee reactions to different teams seems to me to be the most likely explanation.

That does not mean that referees are fixing matches – rather that for some reason they react in different ways to different teams.   But of course if anyone can find a different explanation as to why an Arsenal foul is 78% more likely to be called as a yellow card offence than a Southampton foul, then it would be interesting to hear that.  Especially for a club with someone like James Ward Prowse in the team.


7 Replies to “How a club can commit the most fouls, but get the fewest yellow cards”

  1. Tony

    Ever hear of inverse distance weighting? 🙂 Would you like a map or 8?

    If we assume that England and wales go from -6 to +2 degrees Longitude and from 50 to 56 degrees Latitude, and then we sum over all teams at point (X,Y) the value of one of your 8 columns divided by the distance from (X,Y) to the home stadium of the team.

    I don’t think there is anything profound, you might disagree. But, what format of graph, and do you have a preferred size?

    For distance, I am using the square root of the difference in latitude squared plus the difference in longitude squared. Which distorts the data a little.

  2. I don’t think there is anything profound, you might disagree.

    If I do the math properly (sum of value times weight) divided by (sum of weight), maybe there is a little profound there. I may decided to look at inverse distance squared to see if that offers anything.

  3. Congratulations to the U18 in beating State Aid. The women have sold out their match against the spuds.

  4. Congratulations to the U18 in beating State Aid. The women have sold out their match against the spuds.

    Tony. I went through and made in the inverse square versions. In some instances, the specifics of partitioning changed; the effect of the partitioning WRT geography is more or less the same. I think I avoided creating different files with the same name between the 2 zip archives of PNG files.

    The biggest difference between the two sets of files, is that this second set is boased more towards local data (the first is of a more global nature).

  5. Hmm, blacksheep put the next hread back on topic, need somewhere else to tell a statistical story.

    If we are calculating a statistical surface, the values at the data points do not change. But the changes in the coordinates of all the rest of the data going into the calculations changes in such a correlated way that there is no meaningful difference in the surface calculated (just numerical roundoff error).

    This idea allows us to do something else, if we have enough data. We can partition our data into two parts, both of which are big enough to usefully calculate the surface. The smaller of the two is “just barely” big enough. We fit our surface using the larger data set, and then we calculate a goodness of fit for the smaller data set. We might then look to tune the initial parameters, by producing the “best” goodness of fit on the smaller set. This is known as cross validation (and probably other things).

  6. Words, and the medja.

    In medicine and biology, one can run across the terms ‘in vivo’ and ‘in vitro’. Terms from Latin, meaning of the ‘body’ and in ‘glass’. More or less. Vivo has 4 letters, and so does body. Vitro has 5 letters, and so does glass.

    The above is yet another report on the poor officiating in womens football (but it isn’t just womens football, it extends to anything the 😈 Mike Riley’s influence extends to. Which is nominally all football in England.

    The ‘amateur’ decision talked about in the article, is a late penalty for hand ball was given to a ManU opponent, when the ball hit the head of a defender.

    Hand and head are both 4 letter words which start with ‘h’ and end with ‘d’. To the referee and possibly the medja, that could mean that the two words refer to the same thing. I am sure Dermot the frog could justify the referee’s decision in this matter.

  7. @Tony, shouldn’t your conclusion be made by deductions the article? Looks like your conclusion this time is from another article or from your 160 games article

Leave a Reply

Your email address will not be published. Required fields are marked *