Carlyle stated that “The history of the world is but the biography of great men”, and this theory, with the football manager in the title role, remains ever popular with football fans and pundits. But when it comes to managers there are very few great men. The median tenure of a Premier League manager is less than a season. And now the sacking of Claudio Ranieri poses a serious challenge to the great man theory of football management.
Last March I wrote about the statistical background to Leicester’s extraordinary success. I argued that it had to be understood in the context of an extraordinary season, in particular the fact that the dominant clubs were not as dominant as they usually are. Dominance in football is a function of financial resources. Teams with more expensive squads consistently win more often than teams with less expensive squads, and in football the degree of inequality in spending is very large.
The achievement of Leicester was to win without dominance in financial resources. I said at the time that the chances of this feat being repeated were negligible. I expected them to finish mid-table and although they are in a worse position at the moment, this is still a plausible outcome.
Ranieri was the face of this story, and in our search for a narrative many attributed the success to him. Harvard Business Review even published an article extolling his management skills.
But then this same Ranieri has led a team this season that won only 5 out of 25 games played and was only one point above the relegation zone when he was sacked. This is, therefore, a great moment for all of us to take a deep breath and ask what it is that the data is telling us.
First, there is clearly a great deal of randomness in results. Whatever you think it is that causes winning, it’s subject to huge fluctuations in the short term, with the short term meaning at least 20-30 games. That is partly the nature of football – with so few goals scored it’s always possible for the weaker side to win a single game, and then it is also possible to have a long string of good or bad results.
Leicester winning the title and then being relegated is a very unlikely event, but the probability that over fifty-year one team out of 55 (that is the number of teams that have played in the top division over the last half century) might go through this experience is not so low. Given the probabilities, this could have happened to anyone but will not happen very often (it has happened before- Manchester City won the title in 1937 and were relegated the following season).
If we want to find causality we have to cast the net far wider- to look not just at one team in two seasons. but to look at 100 clubs over 40 years, which is what Thomas Peeters and I have been doing for the last few years .
We find that overwhelmingly it is financial resources which determine success, not the identity of the manager. Because we are working with observational rather than experimental data, there is always a problem of inferring causation from correlation, but we ran a battery of checks which confirm the theoretical prediction.
Indeed, I’ve never seen anyone produce a credible alternative theory to the resource-based theory. Did Manchester City or PSG hire great managers that won titles but then also decide out of the goodness of their hearts to pay the players exorbitant wages? No, they had to hire players in the market whose worth is measured by their ability to win games.
The market for football players is one of the most transparent and competitive markets that one can find. It’s not hard to figure who the best players and teams with the best players usually win. If you have the resources you can buy a winning team, even though in the short term the randomness of results can obscure this truth, once the law of large numbers asserts itself then it clearly becomes the dominant explanatory factor.
That is why in Soccernomics we argued that most managers don’t matter. This is based on the data. A few consistently outperform, but most achieve results that are exactly in line with the resources of the club.
One way to think of this is to think of the contribution of the personnel manager of any large company. Is the success of Apple, or Siemens, or Toyota attributable to the personnel manager? Well, every big company needs one, and there is a long list of tasks which come with the job. For a large company not having a personnel manager would be a mistake. But that’s not the same as being responsible for the overall success of the company.
While it’s clear that in theory the personnel manager could play a dominant role in explaining the success of large corporations, I don’t think many people believe that they do. The personnel manager would not figure very high on the list of explanations for corporate success. And so it should be for football.
Those managers we identify with a successful track record have typically been successful over a long period and usually at many clubs. Ranieri has had a long career without outstanding success in any of his previous clubs. He is a good but not great personnel manager who will be forever remembered for one glorious, extraordinary season.
Many people will take issue with this argument. To begin with, it is a rather cold and unsentimental view of the world, and people like sentiment with their sport. It lacks heart, but that in itself does not make it wrong.
Others might want to take issue with the argument by saying that I have omitted relevant details, or given too much weight to this or that in the argument. So here is a list of inconvenient facts that I think you would have to overcome in order to argue that managers really do matter:
1. In the short term there is a great deal of randomness in results.
2. Over time wages correlate very closely with team performance
3. Causation almost certainly runs from wages to performance, because there is a competitive market for players (many buyers and sellers, easily available information on ability) and so clubs don’t systematically overpay (of course they do in particular cases, by mistake)
4. Average managerial tenure in the Premier League is about two and a quarter years at moment- and if you take Arsene Wenger out the figure falls to about a year and a half. Only 26 out of the 355 managers who have managed clubs in the Premier League since its creation have lasted longer than five years.
5. Given the randomness of results, you probably need at least 3 years in order to identify a significant change attributable to the manager. After all, if you think Ranieri should not have been sacked, it’s because you believe the results this season need to be balanced against the results last season, and so you want him to have a third season in charge to show that he really does make a difference.
6. There is little evidence that success is highly correlated with tenure. Manchester United enjoyed huge success with Ferguson’s extended tenure, but Chelsea and Manchester City have both prospered while regularly hiring and firing managers. And Ranieri himself had not been in the job long before he won.
Very interesting post Stefan. I am curious about if/how your theory of team performance being most tightly related to wages (rather than personnel management or other factors) applies to MLS. I haven’t looked at the numbers carefully, but common intuition is that teams like FC Dallas have managed to have great success in recent years while having among the lowest budgets in the league. Meanwhile some of the big-spending teams (e.g. Toronto) have struggled. Maybe this is simply noise – as exception that proves the rule. But could is relate to the fact that MLS is a more parity-driven league, with the salary cap limiting spending disparities? Or the fact that MLS roster/salary mechanisms make it an inefficient market for talent (e.g. suppressing wages of young players relevant to their open-market value)? Common wisdom seems to that in MLS, the FC Dallas way offers an alternative route to success via good coaching and youth development.
Sorry for not seeing this earlier. There are several posts by me on this website about MLS – the salary cap clearly limits the potential of wage spending to affect outcomes- and when that is the case there are other mechanisms by which teams can improve performance (think college football – there are officially no salaries for players, but some schools are able to field consistently better teams).