Next week is the big Manchester Derby. A match in which victory for either team will strengthen their push for a Top 4 finish. A match which pits arguably the two most respected managers on the planet against each other again. A match which could see £600 MILLION worth of players on the pitch at one time. Perhaps more importantly though, it’s a match which decides bragging rights across a city. Cue a week of build-up in which every ex-footballer and pundit will vehemently claim:

“Form goes out the window on derby day”

The Analysis

“Form” is going to be classified as the amount of points taken over the five games prior to the derby day matchup. There are a number of caveats to this (see the very bottom of this article) but in general it seems a fair, and probably the best, method to use. Data for “Form” will be collected for the Manchester, Merseyside (Liverpool versus Everton), and North London (Arsenal versus Tottenham Hotspur) going back to the 2002-03 season. Only derby matches in which the difference in Form between the two teams is GREATER than three points will be analysed, as clearly it is only in these instances in which one team is more “in-form” than the other. For example, in the first Manchester derby of the 2014/15 season, Manchester United’s Form was 15 points, whilst Manchester City’s Form was six points. This is a difference of nine points and would therefore be included in the analysis. However, in the second Manchester derby of that season, Manchester United’s Form was eight points and Manchester City’s Form was 10 points. This is a difference of two points and would therefore NOT be included in the analysis. Note, matchups were also excluded if they fell so early in the season that one of the teams had not yet played five preceding matches.

The Results

The Manchester Derby

  • Of the 13 eligible cases, Form “went out the window” 5 times. The match “went with the form book” seven times. In one instance the derby matchup was drawn (categorised as “meh” as Form neither went out the window, nor did the matchup go as expected).
  • Since 2003, Manchester City have NEVER gone into the derby matchup with considerably better form (greater than three points difference) than Manchester United.
  • Across ALL derby matches (i.e. even ones in which the Form difference is three or less), Manchester United average 11.6 points prior to derbies they win, 11.5 points prior to derbies they draw, and 10.8 points prior to derbies they lose.
  • Conversely, Manchester City average 7.9 points prior to derbies they win, 7.5 points prior to derbies they draw, and 7.0 points prior to derbies they lose.

 The Merseyside Derby

  • Of the 11 eligible cases, Form “went out the window” just once. The match “went with the form book” 4 times. In 6 instances the derby matchup was drawn (categorised as “meh” as Form neither went out the window, nor did the matchup go as expected).
  • On 3 occasions both Liverpool and Everton have gone into the Merseyside derby with exactly the same Form. Despite this, none of these matches ended in a draw.
  • Across ALL derby matches (i.e. even ones in which the Form difference is 3 or less), Liverpool average 9.3 points prior to derbies they win and 7.9 points prior to derbies they draw (Everton have not won enough to make meaningful inferences regarding losses).
  • Conversely, Everton average 7.7 points prior to derbies they draw, and 7.4 points prior to derbies they lose (again, Liverpool have not lost enough to make meaningful inferences regarding wins).

The North London Derby

  • Of the 12 eligible cases, Form “went out the window” twice. The match “went with the form book” 4 times. In 6 instances the derby matchup was drawn (categorised as “meh” as Form neither went out the window, nor did the matchup go as expected).
  • Arsenal have beaten Tottenham 5-4, 5-2, and 5-2 since 2003. Their Form going into each game was a very mediocre seven, seven, and eight points.
  • Across ALL derby matches (i.e. even ones in which the Form difference is 3 or less), Arsenal average 9.1 points prior to derbies they win, 10.0 points prior to derbies they draw, and 11.2 points prior to derbies they lose.
  • Conversely, Tottenham Hotspur average 9.4 points prior to derbies they win, 7.6 points prior to derbies they draw, and 7.2 points prior to derbies they lose. 

Final Thoughts

Combining the data for all the derby matchups, we find that “Form goes out the window” eight times out of the 36 instances analysed (22%), though interestingly this percentage rises to 38% when looking only at the Manchester Derby. The result “goes with the form book” 15 times (42%); leaving 13 occasions (36%) in which the derby match is drawn, neither going completely as, or completely against, what would be expected based on Form.

Furthermore, Tottenham, Liverpool, Everton, and both Manchester clubs all average more points prior to derby matches they perform better in (i.e. ones they win compared to ones the draw/lose, or ones they draw compared to ones they lose); which is what you’d expect. Only Arsenal work in the complete opposite direction: earning more points prior to derby matches they draw, and even more points prior to derby matches they lose.

Thus, whilst it is a pleasing idea to hold – the thought that the derby matchup is of so much more importance that a player’s increased passion will overcome any notion of good or bad form – the data doesn’t seem to hold up. Teams, generally speaking, perform as expected. Form does NOT go out the window on Derby Day. Which unfortunately, given the form of both teams, isn’t that helpful if you’re wanting to know who to back on Thursday night.

Caveats to the Analysis

  • Form judged on points alone doesn’t take into account opposition played (e.g. a point away at Chelsea would be like a win, whereas a draw at home to Sunderland is like a loss), league situation (e.g. if you only need a draw to stay up or guarantee Europe, or if you have an important cup game coming up), margin of result (e.g. a 4-0 loss is likely more damaging on confidence/moral than a 1-0 loss), or actual results (e.g. five draws may be seen as better than two wins and three losses, though the points are less).
  • Form is based only on league matches and does not include cups. However, given that many teams now rotate players in cup matches (particularly league cup and F.A. cup matches) this is perhaps the best method to use.
  • It could be argued that “meh” (i.e. when the derby matchup is drawn) should also be considered as a case of form going “out the window”, as the in-form team should not have drawn the game. This is a subjective opinion which I would not necessarily agree with.
  • A strong case could be made to weight form such that more recent games are given more value/worth than more distant ones. The decision not to was made purely to keep the analysis simple, though it would perhaps be a better and more accurate reflection of form.
  • From a scientific/statistical perspective, the biggest issue in the analysis is the fact that there is not a “control group”. That is, the analysis does not examine how often form “goes out the window” in non-derby matchups. It could be that form “goes out the window” at a much lower rate than 22% in non-derby matchups, and therefore the cliché is actually rather accurate. In response to this, I believe a strong argument can be made to expect a rate of over 50%, given the prevalence of this cliché, and the assertiveness of those who make it. However, without a control group, we cannot know for sure.
  • Finally, there is the age old sentiment that match results do not always indicate how well a team performed. If a team “didn’t deserve to lose” or “got a lucky win” then the form – and with it any effect it may have on subsequent games – may be an illusion.