Introduction
That richer clubs have a better chance to succeed in the Premier League is almost trivial. The money that these clubs have available to spend on transfers and salaries allows them to sign better players and indeed, more players - football is, after all, a squad game.However, although this is a general rule, there are exceptions. Teams can over- or under-achieve, relative to budget. But the use of terms such as over-achievement require a sense of an expected achievement. But what level of might be expected for a given transfer spend? This post attempts to answer that question and to discover which teams have over- and under-achieved. A related question, regarding the length of time it takes for new signings to "gel" is also examined.
Method - metrics
Using the excellent data available on Transfermarkt, a database of transfer spends has been compiled for each team ever to feature in the Premier League, for every Premier League season (1992-present). Attempts were made to gather data for each team, for each season (whether that team was in the Premier League during that season or not) however, although top-flight data is available going back beyond the start of the Premier League in 1992-3, lower division data dates back only to 2003-4.To ignore the very rapid increase in transfer fees during the sample period (Manchester United broke the British transfer record in 1993 by paying £3.75m; today it stands at £50m), the data is standardised by season using the following formula:
As well as the current season data, "lagged" spends are generated. The one-year lagged spend is simply the total spend one season ago. Lagged spends over the previous 5 years are used, standardised by the teams in the Premier League at the time (so, for example, the mean spend in 2010-11 is not the same as the mean 1 year lagged spend in 2011-12, since the teams in the division in 2010-11 were not the same as those in the division in 2011-12).
The metric used to judge performance is points. Again, this metrics is standardised - it is clear that points distributions change from season to season (75 was enough points for Manchester United to win the league in 1996-97, but would only have been enough for 5th in 2007-8).
The relationship
Following this, a linear relationship is sought between the standardised points metric as the dependent variable and the following independent variables:- Standardised transfer fees paid (current season)
- Standardised transfer fees paid (current season - 1)
- Standardised transfer fees paid (current season - 2)
- Standardised transfer fees paid (current season - 3)
- Standardised transfer fees paid (current season - 4)
- Standardised transfer fees paid (current season - 5)
It is interesting that, towards the extremes of actual values, the expected values are in general not as extreme. This suggests that the distribution of points in the league table is different to the distribution of transfer fees paid - that the marginal value of transfer spend is higher at either end of the table.
The coefficients of the relationship are plotted below:
It can be seen that the largest coefficient is that for the current season spend. This value decreases as the lagged number of seasons increases - up to season 4, at which point the coefficient jumps upwards. This suggests that instant rewards should be seen for money spent which is not always the accepted view.
Results
A time series of a club's performance relative to expectations can be generated. A few are picked out below:Arsenal
- Consistent overachievement
- Expectations over 4 or the last 5 seasons, given the transfer fees paid, have been for a below-average number of points - Arsenal's transfer outlay has been consistently below average.
Chelsea
- Roman Abramovic bought the club in 2003
- Jose Mourinho's first season was 2004-5
Everton
- David Moyes became manager at Everton in 2002
Liverpool
- Rafa Benitez was manager from 2004-10
- Kenny Dalglish was manager throughout last season
Manchester City
- Despite winning the league last season, Manchester City only slightly exceeded expectations
Manchester United
- Sir Alex Ferguson has been manager throughout
- Consistently exceeded expectations
Tottenham Hotspur
- Harry Redknapp was manager from 2008-12
Conclusion
This model allows the performance expectations of a team to be quantified. From this, performance relative to these expectations can be shown. I am pleased that the quantitative results shown here correspond fairly well with my own quantitative preconceptions, though it must be noted that, in recents seasons, particularly high transfer spends by the likes of Chelsea and Manchester City have led to the average transfer spend in any given season being inflated: for example, in 2010-1, only 4 teams, Aston Villa(!), Chelsea, Liverpool and Manchester City had spends which were above the division average.The following conclusions have been drawn in this post:
- Expecting immediate return from transfer spend is realistic
- Sir Alex Ferguson has consistently overachieved
- David Moyes has performed very well over the last several seasons
- Rafael Benitez did underachieve in his final season, but subsequent managers have not improved the situation
Further work
This work was designed to investigate performance relative to budget. The following transfer-based studies can also be carried out using similar data:- Performance relative to net spend
- Quality vs quantity - is it preferable to buy a small number of top players or a larger number of less good players?
Please let me know what you think of this post, either in the comments below or via Twitter: @hpstats
No comments:
Post a Comment