Tuesday, 26 July 2011

A question of experience

When we appointed Chris Powell as manager the New York Addick (who seems to have gone into purdah for the summer) argued that we should have appointed an 'experienced' manager.

So what is the evidence for the benefits of experience? I have been looking at the book on Football Management by my colleague Sue Bridgewater who runs courses for the PFA and the LMA. She argues that experience should be considered as a factor in football success, but the evidence on its importance appears to be a little mixed.

There is certainly a significant difference between the number of years experience and the win percentage of managers. Those with no previous experience have a ratio of 33 per cent wins, but this goes up to 45 per cent for those with ten years or more experience.

Sue notes, 'Experienced managers have learned lessons through time, perhaps through making and learning from mistakes. They are also the survivors, those who through natural ability and learning over time have a proven ability to do the job.'

However, experience matters much more in the Premier League is far more important than for managers in lower leagues. Not surprisingly Premiership managers have more experience:

Premier League 9.6 years
Championship 6.5 years
League 1 4.8 years
League 2 3.6 years

Sue suggests, 'Managing star players in the full glare of the media spotlight is not easy and it may be for this reasons that managers who have been there and seen it are better able to manage at this level. There is something distinctive about managing these strong egos and management experience does help with this.'


Floyd said...

I am waiting for a stunning conclusion Wynn. So, with SCP's 0.6 years versus the average of 4.8 years what do we conclude - we don't exactly have big egos to match those of Super Mario Balotelli or young Cashley do we.

What should SCP's win ratio be at the end of the next season when he 'may' have 1.6 years experience...

Chris said...

There are two reasons for preferring a Manager with experience to one without.

The first is the expectation that experience helps; Managers learn with experience and become more effective with time. Your suggestion, I assume, is that at League One level the evidence for this effect is ambiguous. In fact, the hypothesis is very hard to test. I haven’t had the benefit of reading your colleague’s work [do you have a reference?], but even her Premier League result might be the simple result of survivorship bias. There is also the significant effect of outliers [Ferguson, Wenger] and a large number of exogenous variables that are very hard to control for. I’d stick with simple logic. All other things being equal, experience must help. How many other “skill based activities” can you name where it doesn’t?

The second reason for preferring a Manager with experience is compelling; experience provides concrete evidence of performance. Let’s suppose that Managers can be equally divided into three types; Good, Bad and Indifferent. Controlling for all other possible selection criteria [which should apply equally to all three types], then we can say that if a Manager with no experience is chosen there is only a one in three that he will be Good and an equal chance he’ll be Bad. However, if we select from experienced Managers, i.e. with track records, we have the chance to improve these odds because know how they’ve fared in their careers so far. We have no certainty, course, but had Charlton hired an Eddie Howe [I know he turned down the job], a Darren Ferguson or a Keith Hill, then it would have been reasonable to suppose that the odds of the Manager being “Good” would have been better than one in three and hence better than for a Manager with no experience.

Of course, it is possible that Chris Powell scored so highly on the “other selection criteria “ that his inexperience didn’t matter, but it is perhaps a little surprising that this could have emerged in an interview lasting less than an hour. In short, there really is no obvious reason why you would choose a Manager with no experience when Managers with experience and positive track records were available.

I’d imagine this was NYA’s point don’t you? It was a surprising appointment by any standards and one that is yet to be justified by results. Fingers crossed for the forthcoming season.

Wyn Grant said...

Thanks for the full reply. The reference is Sue Bridgewater, Football Management, published by Palgrave-Macmillan. Survivorship bias may well be an issue, although I think she is aware of that. Of course, experience counts, although I think one of the problems with football management is specifying what the requisite skills mix is which, of course, makes the hypothesis even harder to test. There are also plenty of experienced managers who do not succeed.

Chris said...

Wyn - Many thanks for the reference.

I think you are right about the skill mix question.

There are clearly Managers who have [or had] superior skill based on their results. Neil Warnock is an example of a Manager whose record is so impressive that, statistically speaking, it is extremely unlikely that he doesn't possess very superior skill. But what, exactly, is that skill?

Having read the Brian Clough biography by Nottingham Post journalist Duncan Hamilton [who was very close to Clough] I was left bemused; just what was it that enabled a quite dysfunctional human being to be so effective, over a prolonged pereiod, as a Football Manager?

One can imagine a selection panel setting out the things it was looking for in a series of interviews for a Manager vacancy; 1) must be a complete jerk, 2) with an uncontrollable temper, 3) must be able to name at least half the players from the team he last managed....

To be more serious, it really isn't obvious what you'd look for.


Wyn Grant said...

Clough in a way is the exception that proves the rule, he was so way out in left field. But he did have that indefinable ability to inspire players. Contextual factors can help a manager, especially a good relationship with a supportive board, e.g., Murray/Curbishley. Ultimately I think that the skill mix is indefinable, although one can identify the elements.