Manchester United, Poor Decision Making and the Problem of Small Sample Sizes

Manchester United are one of the most successful clubs in English football (soccer) history[i] and also the second most valuable sports teams in the world[ii].  It is therefore somewhat puzzling that they made a clear and unambiguous error of judgement in their recent decision to appoint former player Ole Gunnar Solskjaer as their permanent manager*.  This is not a slight on Solskjaer nor his abilities as a manager (upon which I have no strong view), but rather a critique of how such an esteemed and well-resourced organisation could apparently fall victim to statistical naivety and a number of behavioural biases.

Following the historically successful tenure of Alex Ferguson, Manchester United went into a period of relative decline moving through a succession of managers culminating in the sacking of Jose Mourinho midway through an underwhelming 2018/2019 season.  In what appeared to be a prudent course of action; rather than make a rash, reactionary managerial appointment the board decided to appoint a temporary manager and then make a decision on a permanent appointment at the end of the season – thus affording them more time and a likely greater opportunity set of higher calibre managers.

The temporary manager appointed was Ole Gunnar Solskjaer.  Whilst Solskjaer had little managerial pedigree outside of a disappointing spell in the English Premier League and a longer tenure in the minor Norwegian league; he was a cult hero as a Manchester United player – known for his goal scoring ability as a substitute and steeped in the history of one of the club’s most successful periods. As a short-term ameliorative it seemed to be a sensible choice.  Few expected his tenure to run past the end of the season.

The form of the team improved immediately following Solskjaer’s appointment as Manchester United went on to win 14 of his first 19 games in charge.  Despite his experience seeming entirely unsuited to a job of such magnitude, this exceptional run led to an increasing clamour for him to be offered the managerial job on a permanent basis.  With at least 10 games remaining until the end of the season, Manchester United decided they had seen enough and offered him a three year contract.

At face value the move was understandable – Solskjaer’s impact had transformed the fortunes of the team and this had provided compelling evidence of his skill as a manager – but was this really the case?  Were 19 games enough to make such a judgement? And what was the benefit of making a decision before the end of the season?

The underlying inference of Manchester United’s decision to make Solskjaer’s appointment permanent was that there was a causal link between Solskjaer’s skill as a manager and the upturn in the team’s fortunes.  Although this is the instinctive conclusion to draw it is not necessarily the correct one – there are a range of other confounding factors that could have contributed to improved form:

Simple mean reversion: It is likely that a manager will be sacked after an unusually poor run of form – when their recent results are some way below average.  The supposed influence of Solskjaer may just have been reversion to the mean.

A run of easier games: Poor form and a managerial change can often coincide with a run of more difficult games against tougher opponents with a resurgence arriving when the schedule of matches becomes easier.  Mistaking skill for a spell of less demanding games is a major risk if you are working with a small sample.

Players trying harder: A short-term upturn in performance may simply be a reaction from players who were disaffected under the struggling former manager.  They may expend more effort and attempt to impress the new manager – this is a temporary phenomenon and no reflection of managerial skill.

Sheer Luck:  Although hard to accept, when dealing with small samples outcomes may simply be a result of luck and randomness.

The most grievous element of Manchester United’s decision making was their explicit choice to reject the opportunity to observe a larger sample of evidence (more matches), particularly given Solskjaer’s lack of experience at the highest levels of management.  Rather than wait until the end of season they rushed to appoint Solskjaer despite their being at least 10 games remaining, and therefore left themselves beholden to the vagaries of a very small sample of evidence.  A c.50% increase in sample size would not have made the evidence infallible, but the acquisition or more relevant information was both beneficial and costless for Manchester United in this situation – there was no downside to waiting and learning.

The only possible reason for rushed permanent appointment would have been if there was a risk of a Solskjaer accepting a job at another club so they needed to secure his services – however, given his lack of pedigree and affinity with Manchester United this was never a genuine threat.  The club therefore opted to make a decision on a small biased sample despite having a free option to materially increase the amount of available evidence.  It would be unfair to suggest that Manchester United were alone in this thinking; there was seemingly a growing sense from fans and interested observers that Solskjaer’s strong early run had seem him rightfully earn the position on a permanent basis.

Dealing with small samples of evidence allows our ingrained behavioural biases to run amok as we seek to draw meaning from incredibly noisy data.  Not only were Manchester United suffering from a bout of outcome bias and myopia, they also succumbed to the irresistible lure of a compelling narrative.  Solskjaer’s playing career and indeed his words as manager harked back to the glory days of the club that had been somewhat lost in recent years.  The combination of this nostalgic yarn and a short period of strong results simply proved too intoxicating.

With somewhat grim inevitability the form of the team has deteriorated since the permanent appointment was confirmed and it is now doubtful whether the club would have made the same decision had they been in possession of this additional information.  The subsequent performance of Manchester United under the stewardship of Solskjaer is however something of an irrelevance when we come to assess the quality of the decision to make him the manager on a permanent basis.  Even if Solskjaer goes on to achieve great success at the club, it does not change the fact that Manchester United willingly and needlessly made a poor decision based on an inadequate sample of evidence.

I have commented on this blog previously that there is no better place to observe our behavioural foibles in full bloom than financial markets, but I should add sport alongside finance –  it is also provides wonderfully fertile ground for poor decision making and biased judgments from both participants and observers.  Of course, many of the traits exhibited by Manchester United in their managerial appointment will be painfully familiar to investors – the short-term thinking, the confusion of randomness and skill, the danger of small sample sizes and the lure of a good story.  Seeing one of the largest and most prominent sports teams in the world fall victim to such common decision making problems is both worrying and comforting.

* A manager in English football is analogous to a coach in US sports.





2 thoughts on “Manchester United, Poor Decision Making and the Problem of Small Sample Sizes

  1. Pingback: Top clicks this week on Abnormal Returns – The Wealth Builderz

  2. Pingback: Investors can learn from Manchester United’s ‘statistical naivety’ – The Irish Times – Manchester United

Comments are closed.