Manchester United: an analysis of the spring kings
Federico Macheda scored twice in April 2009 to seize momentum for Man United in the title race, providing more evidence that Man United excel in the springtime (Flickr: miamiking22)
Manchester United look set to win their 19th league title this season, eclipsing Liverpool’s championship record.
The achievements of Man United over the last 18 years are the envy of every club. Such has been their dominance, it borders on the unexplained. How can a club accomplish such consistent success without ever coming close to being knocked off their perch?
The common denominator is of course Sir Alex Ferguson. Much has been written on his man-management, tactical and transfer market skills, all critical towards Manchester United’s success. One theory is that Ferguson knows how to get it right at the important times of the season, namely in the spring months at the end of each campaign. This is a theory he has fuelled himself in the past:
I keep saying it and nobody seems to listen, but March and April are the months that decide the league title. A lot can happen in those months. We have lost leagues in March and April in the past and we have also won them by going on fantastic runs at that time.
I’ve previously written at length about title-race momentum, and concluded that it tends to swing critically in April; right at the heart of the spring period. I also noted that Manchester United have benefitted from these April ‘swings’ more often than not, giving credence to Ferguson’s belief.
I’ve chosen to revisit this topic to try and measure what makes Manchester United so good during April and March. To do this, I compiled information on Manchester United’s results from the last 9 completed league seasons (2001/02 to 2009/10). United won the title in four of these seasons: 2002/03, 2006/07, 2007/08 and 2008/09.
The difference in win percentages for United over the course of the season was immediately striking. Defining pre-winter as the months from August to November, winter as December to February, and spring as March to May, this table shows how the Red Devils have fared in the stated periods:
|
Win % |
All Seasons |
Non-title winning seasons |
Title-winning seasons |
|
All Matches (All months) |
67.0 |
63.7 |
71.1 |
|
Pre-Winter (Aug, Sep, Oct, Nov) |
60.6 |
59.4 |
62.1 |
|
Winter (Dec, Jan, Feb) |
73.9 |
70.6 |
78.4 |
|
Spring (Mar, Apr, May) |
66.7 |
60.4 |
74.4 |
From these raw statistics, it appears United’s form typically experiences an n-shaped curve over the course of the season. They’re notoriously slow starters, but peak in winter, before (relatively) falling away in the final three months of the season.
This doesn’t immediately mean that Manchester United are better in winter than in spring, or better in spring than pre-winter. These numbers may be biased; the fixture computer may have consistently given United a disproportionate number of difficult away matches in the early part of the season, and a high number of easy home matches in the middle part of the season.
To allow for these potential biases, I use a logistic regression to model the win percentages of Manchester United at different stages of the season. Essentially, the model can take into account the location of United’s match (home or away) and the strength of the opposition whilst testing if they are better in spring than parts of the season.
The opposition’s strength is proxied by their final league position. Although this isn’t the best measure since it doesn’t account for the form of United’s opponent at the time, the results show United’s results are highly correlated with their opponent’s strength using this method.
Thus, taking into account the match location and opposition strength, the key results I found are as follows:
- In seasons where Manchester United won the Premier League, they were 15% more likely to win in winter compared to pre-winter.
- They were also 13% more likely to win in spring than in pre-winter in these seasons.
- However, there was no evidence to suggest they are more likely to win in spring than in winter, or vice versa.
- In seasons where Manchester United did not win the Premier League, there was no difference in win percentages across difference parts of the season.
- I also found no evidence to suggest United score more late goals (defined as from the 80th minute onwards) in spring or winter compared to other stages of the season.
Whilst the results confirm the belief that the spring period is important, they also highlight the significance of getting results in the winter period: results in winter are just as important as results in spring.
This point is confirmed by looking at the five seasons where United have not won the title. In all of these seasons, there is no significant difference between win percentages in winter compared to pre-winter, despite the sizable difference on the table above.
What this means is that in these seasons Manchester United have failed to sufficiently make up for their slow start that they typically suffer. In this respect, they didn’t lose these league titles in spring; they’d already lost them in winter. If they don’t have a good enough winter, they won’t have a good enough spring. This refutes Ferguson’s earlier claim that United have lost titles in March and April, or at least in the period from 2001/02 to 2009/10.
The results give us some insight into why United have been so good in the months of March and April in title-winning seasons – they’ve already built momentum during the three months prior. In the seasons where they did not do this, they invariably did not win the league.
It isn’t a case of Ferguson’s team flicking a switch, suddenly aware that it is “squeaky bum time.” Instead, it’s a case of slowly building towards matches where pressure and points differences become more obvious. Results in April appear to have a more pronounced effect on the league table due to the scarcity of matches left, so we remember these matches more than the equally important results during the winter period.
The evidence extends to the current season. Before December, United had won 8 out of 15 matches, a win percentage of only 53%. During the December-February period, they won 9 out of 12 matches, or 75% of matches. The difference in win percentages is 22%, and although I’ve not accounted for bias, it’s fair to suggest there’s been a significant increase in United’s win percentage.
The signs were obvious by the end of February; the pattern of the current season mirrors how Manchester United have won titles in the past.
Despite poor results at Stamford Bridge and Anfield, United’s winter form has built the momentum and confidence to come from behind as they did at West Ham, or grind out results as they did at home to Bolton.
You’ll note that in the ‘key results’ I found that United do not have a tendency to score more late goals at any particular stage of the season. I was curious to see if their ‘never-say-die attitude’ translated into scoring late goals at so-called important stages of the season; this was proven not to be the case.
My analysis is far from exhaustive, and only scrapes the surface as to why United finish seasons so strongly. It also doesn’t include comparisons with other teams; this is merely a look at Manchester United’s trends.
However, I believe it does give insight into the formula of Manchester United’s success. For United to be spring kings, they need to be winter… warriors. It is their form during the months of December, January and February that determine whether they will perform during March, April and May. The winter months are just as, if not more important than the spring months.
It doesn’t make much fun to know the title winner by the end of February, but more often than not you can tell if it’ll be United’s year or not.
How big an influence do you think United’s winter form is on their success? Why do they appear to be so good during the spring period? Please provide your thoughts in the comments section below.
5 Added Minutes 
Hey, great analysis.
Perhaps the idea that United are so much better in the spring is simply to do with media hype – journalists seem to love a cliche that they can revert back to season after season. This doesn’t mean that United are poor in spring, but I recently found this information courtesy of sky:
Points gained in the last 10 games of the season:
2009/10
Arsenal: 17, Chelsea: 25, Man Utd: 25
2008/09
Arsenal: 23, Chelsea: 25, Man Utd: 25
2007/08
Arsenal: 18, Chelsea: 24*, Man Utd: 23
2006/07
Arsenal: 13, Chelsea: 20*, Man Utd: 20
2005/06
Arsenal: 23, Chelsea: 19, Man Utd: 23
2004/05
Arsenal: 25, Chelsea: 24*, Man Utd: 15
2003/04
Arsenal: 20*, Chelsea: 18, Man Utd: 17
* Unbeaten
OVERALL
Arsenal: 139 (an average of 19.9 points in the final 10 games)
Chelsea: 155 (an average of 22.1 points the final 10 games)
Man Utd: 148 (an average of 21.1 points in the final 10 games)
The spring period of March onwards generally contains 9 or 10 matches for each club, so roughly speaking this is a fairly good guide to this period.
As it shows, United have never gained more points in the Spring than both their rivals (though in certain years, such as last season, both they and chelsea gained the same points). In fact, in 2008, Liverpool, who finished 2nd but were omitted from Sky’s list, gained 3 more points that United even though Fergie’s team won the title. Likewise, in 07/08 they gained less points than chelsea in the spring who finished just two points behind them.
This would again back up your idea that man u win the league in winter rather than spring, while showing the idea that Man U are better towards the latter end of the season as an overused remark, not substantiated by evidence. Perhaps this is something to do with the fading away of Arsenal in 3 of the last 4 seasons which, by comparison to United, has helped perpetuate the myth. But by the same token, Chelsea’s spring results should also look better, and nobody mentions their run-in.
Of course Alex Ferguson will wish to highlight the importance of the spring period, his players are sure to gain confidence from the media constantly repeating their ability to win games come the end of the season compared to their rivals, even though the facts don’t back it up
Brilliant post, now I don’t need to do a comparison to the other teams! Some really interesting numbers there, I really didn’t think United’s lack of dominance in spring would be so pronounced.
Fascinating stuff, as always. You seem to be confirming the notion that three points picked up at the start of the season are just as valuable as three points picked up at the end of the season – which may seem like a duh! comment, but it runs counter to another notion that ‘winners’ step it up in when it comes to the crunch and that we should put more weight on those matches in trying to define why they won.
As an aside, as far as I’m concerned the defining factor in Man Utd’s success is Alex Ferguson. As evidence I offer his record in Scotland. Since Kilmarnock won the Scottish League in 1965, the Old Firm have only failed to win the title on four occasions – and three of those occasions have been by Aberdeen, managed by . . . Alex Ferguson. Kills me to say it, but he is the greatest manager of them all
Well I think the model shows that results at the start of the season aren’t as important to Manchester United but may or may not be as important to other clubs (I don’t know, I haven’t analysed their results).
But yes, totally agree on Sir Alex Ferguson.
I know this post is related is based on stats but there are a lot of other factors involved in winning a title and even though you have won a lot in the winter months the spring fixtures are the ones which have more pressure involved and teams that dont handle them well will lose even though they have won consistently in the winter months. Take a look at the first table you posted –
Non-title winning seasons –
Overall win % drops by 3.3% from avg
winter months win % drop is 3.5% and
spring months win % drop is 6.3%
Title winning seasons –
Overall win % increases by 4.1%
winter months win % increases by 4.5% and
spring months win % increases by 7.7%
You can clearly see that in title winning and non title winning seasons performance in spring months deviates more from the average
I’m not sure you understand the model, or my point.
Those raw percentages are fine to look at – but they could be affected by biases.
It could be that Manchester United generally receive easier games in spring in seasons that they’ve won the title, and tougher games when they haven’t won the title.
Hypothetically, say in 2007, 2008 and 2009, United had run ins of (generally speaking):
Stoke (h), Wolves (a), Fulham (h), Bolton (a) and Wigan (h)
But in the seasons they didn’t win the title, they had run ins of:
Chelsea (h), Tottenham (a), Arsenal (h), Liverpool (a) and Man City (h)
Clearly, United are much more likely to perform in the first run in. Of course, those are the extremes, but instead of looking at each season, case-by-case, to see if the run in was ‘difficult’ or not, I can model all their results using a logistic regression to save me time and give me better results.
I model the results taking into account the location (home/away) of the match, and the opposition strength.
And by doing this it shows that United’s winter form is significantly better than their early season form. HOWEVER, the model also shows that their spring form is significantly better than their early season form.
So, as I’ve said, their winter form is equally important as spring form.
This is an oft-neglected fact. I never said their spring form was not important – I’d be a fool to suggest that.
This is also purely a study of Manchester United, not a comparison with other teams – I’m not looking at whether other teams can handle pressure or not. That’s a separate study in itself.