Data: the impact of past performance on future results
We examine data from the last five seasons to discuss how United’s performance in the Premier League is affected by the result of the team’s previous match. All Premier League games are included provided that the previous match was played in a competitive competition such as the Premier League, FA Cup, League Cup, Europa League or Champions League. Table 1, below, shows United’s win percentage in the league based on the team’s previous match result from 2014/15 to 2017/18.
Table 1 shows that United win nine percentage points more games in the Premier League following a loss than following a win. This indicates that the bounce back effect from a loss may be greater than the confidence boost following a win.
As an interesting aside, the difference shown in Table 1 is most pronounced for the season under David Moyes. The table shows that staggeringly, under Moyes, the Reds were almost twice as likely to win if the team had lost the previous match than if it had been won. A possible explanation for this is that Moyes struggles to develop a winning mentality at clubs, but does have the ability to get a team pumped following a defeat. This might explain why he struggled to manage at a bigger club where winning consistently is much more important. The figures in Table 2, below, show the Premier League win percentages under Moyes.
Although these results are interesting, they do not necessarily mean that the result of the previous match is causing the differences in win percentage. There may be other factors involved.
For example, matches are typically scheduled so that teams play a home match and then an away match. As teams are more likely to lose away from home it makes sense that a team is more likely to win following a loss simply because they are more likely to be playing at home. The next step is therefore to control for these interfering effects to see if the impact that the result of the previous match has can be determined.
Controls are included for match location – home or away – and for the quality of the opposition based on league position, average goals scored, average goals conceded, and form over the previous five matches. Therefore, Table 3, below, shows the results of this analysis – the impact of various factors on the likelihood of United winning a Premier League match.
Table 3 shows that United’s chance of winning a Premier League match is increased by a statistically significant 16 percentage points if the team has lost the previous match rather than won it. Given that the main interfering factors, such as match location and opposition quality, have been controlled for, we can say with reasonable confidence that the Reds are more likely to win when they have lost the previous match than when they have won it.
With this, and Sunday’s Manchester derby in mind, United’s two-minute double against CSKA no longer seems to be such a positive. Mourinho’s men may in fact have shot themselves in the foot with two deft finishes.
That said, the analysis above does not examine specific player level effects and the benefit to Lukaku’s confidence may be very important at this moment. With City suffering defeat to Shakhtar on Wednesday, it will be interesting to see how the game on Sunday plays out.