Everton detest Liverpool and vice-versa.  Spurs hate Arsenal. No-one likes Chelsea. Everyone hates Leeds.

In my last post I looked at the structure of rivalries in the top flight of English football.  In this post, I look at what happens when teams meet their arch rivals in the league.

Are matches between bitter foes different to matches between non-rivals? A quick scan of the basic numbers for the last ten seasons would certainly say so.

  • Liverpool/Everton games in this period produced a total of 10 red cards. This was only equaled by  Liverpool /Newcastle match-ups  and is more than three times the number produced by the average Premiership  game.
  • Chelsea/Manchester United fixtures garnered the most yellow cards (109) with Arsenal/Manchester United (97) and Chelsea/Liverpool (94) not far behind.   Yet over the same period, the average Premiership clash produced only 64 yellows.
  • The fixture producing the most fouls in the period was   Liverpool/Everton with  560.  Manchester United/Liverpool and Chelsea/Arsenal were close with 545 and 544 respectively, while the average match produced 460.

But to draw any general conclusions about rivalries, we need a more systematic analysis.

The first problem is knowing which teams are rivals and which are not. Sports writers have offered various lists based on their personal opinions, but the best data comes from Chris Whiting’s 2012 survey of football fans. Chris asked fans to identify their club’s main and secondary rivals, and based on his data, we can define three types of fixture.

  • Non Rival (R0) fixtures are matches where neither set of fans consider the opposition team to be a rival club.
  • Single Rival (R1) fixtures are matches where one of the clubs considers the opposition to be their rival but not vice-versa. For example,  Manchester City fans consider Liverpool to be a rival club, but Liverpool fans don’t consider Manchester City to be a rival club. So Liverpool /Manchester City fixtures are R1 fixtures.
  • Double Rival  (R2) fixtures are matches where both sets of fans consider the opposition to be a rival club. (For example, Liverpool/Everton.)

The crucial question is whether the team performance indicators (such as passes, shots, tackles etc.) in R1 and Rfixtures differ from those for R0 fixtures.

To illustrate the method, consider tackles in the Liverpool/Everton rivalry. In each season we can calculate the average number of tackles for both teams when they are playing R0 fixtures (i.e against non-rivals). Then, the number of tackles we’d expect to find when the teams play each other is simply the sum of the two averages.  If we find consistently more or consistently fewer tackles than this, we can conclude that playing against your rival makes a difference on the pitch.

I used data from  eight seasons of the Premier League, and around 1100 Championship matches. I did the analysis for all the  R1 and R2   fixtures in the dataset. (The R1 fixtures turned out to be almost statistically indistinguishable from the R0 fixtures, so I’m not reporting them.)

The Results

I divided the analysis into four domains;

  • Competing
  • Discipline
  • Footballing skills
  • Emotion


R2 Difference
Performance Indicator Average per R0 fixture +/- % change
Tackles 41 +4.36 10.7% ***
Fouls 23 +1.91 8.4% ***

The average column shows the average volume of tackles and fouls for R0 fixtures.  The +/- and % change columns show the raw difference and percentage changes for the R fixtures respectively, and the stars indicate the statistical significance of the difference.

As we might have predicted, tackles and fouls are significantly up.  The data suggests players are competing harder in R2  matches, and are more eager than usual to regain the ball when out of possession.


R2 Difference
Performance Indicator Average per R0 fixture +/- % change
Yellow Cards 3.1 +0.99 32.1% ***
Red Card 0.16 +0.11 67.5% ***

Closely related to competitiveness is discipline. On average an R2  match attract one yellow card above expectation.  The probability of getting a red card is also substantially raised, and although the actual numbers are small, the increase does indicate a significant deterioration of discipline under the surface.

 Footballing skills

R2 Difference

Performance Indicator Average per R0 fixture +/- % change
Goals 2.6 -0.28 -11.0%*
Shots 20.8 -2.20 -10.5%***
Passes 792 -79.65 -10.1%***
Pass Completion 71% -2.10% -2.9%***
 * p <=.05, ** p <= .01, *** p <=.001

In a match between rivals, there is less football than you might expect;  Goals, shots and passes are down by about 10% and there is a small but significant drop in pass completion.

I looked a few other indicators as well, which I left out of the table to avoid clutter.  I thought that there might be more activity high up the pitch in R matches, but it turned out not to be the case. The percentage reduction in passing was uniform, around 10% in each third of the field.  We don’t see any difference in the balance of activity across the pitch.  I also wondered if players might be inclined to dribble more instead of passing, but again I found nothing. No difference at all.  Finally, perhaps there were more long passes as hyped-up players reverted to more primitive tactics? No again.  It looks as if the game is being played at a more deliberate pace which perhaps indicates that the high tension slows down decision-making.


Digging deeper into the stats (on a smaller dataset of four seasons), I found there were three types of cards that were more frequently shown in matches between rivals. I also found a 15% increase in injuries and a big increase in the number of errors.

R2 Difference
Performance Indicator Average per R0 fixture +/- % change
Card: Argument 0.09 +0.10 112% *
Card: Excessive Crowd celebration 0.05 +0.08 158% **
Card: Dissent 0.27 +0.12 46% *
Injury 3.13 +0.46 15%*
Error Leading to Goal 0.44 +0.21 47% ***

Of course the raw numbers of cards are quite small, but their scale is irrelevant. They are the visible indicators of an underlying level of emotional intensity on the pitch, and measure the tip of an iceberg.  The absolute size of the tip doesn’t matter, what we are really interested in is the iceberg underneath. - and as the percentage changes show, this is massively heightened in  R2 matches.

We also see an increase in injuries and a large increase in errors suggesting that the psychological pressure of playing against a strong rival can substantially impair a players judgment.

The Take-Away

There is strong evidence that matches between rivals play out differently than normal matches. It is not only the fans that are affected.

The reduced volumes of passing and shooting suggests that player decision-making is slowed down. Yet other aspects of the game are intensified. There is a heightened passion that increases mistakes, and an uptick in tackling and fouls that spills over into disciplinary offences, and more frequent injuries than in ordinary matches.  Altogether It seems that In in these intensely fought contests the pressure of the occasion impels players to kick the ball less and kick each other more.