Managed decline

When numbers get serious they leave a mark on your door

I’ve found a brilliant coin to play heads-or-tails with: it just doesn’t know when it’s beaten. I first used it the other day, and although it lost the first time, it went on to win best-out-of-three. It’s definitely worth keeping.

That’s obviously absurd logic, isn’t it? And yet, we see the same logic playing out again and again. Last week I talked about the human tendency to impose a narrative on random patterns in opinion polling data, and this week I’m going to argue that a similar thing happens with sport and with life in general, not helped by the recent Moneyball-inspired tendency to quantify everything in sight.

It’s ironic, because I’m a numbers person. I’ve written a whole book about the value of data in making sense of the world¹. But part of the aim of Numbercrunch was to argue against spurious precision in reporting data, and to argue that luck plays a significant role in everyday life.

So you can imagine how I felt when I saw the following table on the BBC website, in an article about who should take over from Jurgen Klopp as manager of Liverpool FC:

The whole thing just feels absurd: it’s Top Trumps gone mad. Even for a tangible physical quantity like temperature, we generally report forecasts to the nearest degree. We might report record temperatures to the tenth of a degree, following careful investigation and calibration. So the idea that you can measure “attacking coefficient” (whatever that is) to four significant figures, and therefore rank Thomas Tuchel ahead of Julian Nagelsmann by 0.03 of a unit (a 0.15% margin!) seems patently nonsensical.

Indeed, this kind of spurious precision should in itself give us serious doubts. It feels like someone reporting numbers without thinking about what they mean. It’s likely the qualities identified are important, and perhaps you might give rough numerical scores to start a discussion, but it seems unlikely to me that you would gain useful insight by doing this on anything finer grained than a five- or ten-point scale.

But beyond that, the whole thing neglects the role of luck, of regression towards the mean. For example, the whole analysis concludes that by some margin the best candidate to replace Klopp is Ruben Amorim of Sporting Lisbon. And Amorim’s win percentage is indeed impressive: but 5 years ago he wasn’t even a manager - his list of games managed isn’t a huge sample size!

Similarly, Xabi Alonso is odds-on favourite to be the Liverpool manager, based on an impressive 18 months at Bayer Leverkusen as well as the intangible presence of “Liverpool DNA”. But you could have made a similar argument for Steven Gerrard’s managerial destiny in autumn 2021 - and as an Aston Villa fan you don’t need to tell me how that worked out.

Or take Thomas Tuchel. Is he the manager who did the quadruple and reached two successive Champions League finals (winning one)? Or is he the manager who was fired from Chelsea after a poor start to the season, only won the Bundesliga with Bayern Munich thanks to Dortmund collapsing on the last day of last season, and is in serious danger of being at the helm when they don’t win the league for the first time in over a decade?

And the answer is, of course, he’s both. Just as with the silly example of the coin toss, and with the opinion polls I wrote about before, it’s always tempting to build a narrative and to ascribe causality to essentially random events. But football and other sports are inherently unpredictable, which is of course the whole joy of them. There are always days when the better team just don’t score. A key player can be suddenly injured, or suspended, or have issues in their private life that affect their form.

Even strong teams don’t win every game, and just by random chance sometimes they will undergo a run of three or four games without a win, putting a manager under media pressure. For this reason, while there is evidence that changing manager can somewhat improve things (these authors estimate a five point per season gain from changing manager for example), because sackings usually take place following a run of poor form you can’t ignore the possibility that things might have got better anyway by random chance alone. Further, it’s not necessarily clear that the money it takes to pay off an unsuccessful manager and recruit an incoming one could not have been better spent on players.

Overall, it feels extremely reductive to tell a story of The Great Man at the helm who can control events at will. Even a manager like Klopp has had ups and downs. Successive yearly points totals of 75, 97, 99, 69, 92 and 67 tell a story of fluctuations, rebuilds and shifting form and tactics. Perhaps his legacy will come through his final season, where a quadruple is still possible. But even just the first leg of this, today’s League Cup final, is close to a toss up. Sure, Liverpool are favourites, but the Betfair Exchange odds imply they have a roughly 60% chance to Chelsea’s 40%. So it’s not quite a coin toss, but when tomorrow’s newspapers are writing a narrative of either Klopp’s genius or Pochettino’s rehabilitation, it would be sensible to remember that things could easily have gone the other way.

Indeed, there are lessons for all of us here. Whatever your job, it’s likely that there will be good and bad runs over the course of your career due to chance alone. If you work in sales, there will be times when it feels like nothing is working, apparently for no good reason. If you are a lawyer or a doctor, then sometimes tricky cases or bad decisions will feel like they are clustering together. There are classic books which were rejected hundreds of times by publishers.

Even in science, where we like to pretend that truth is always obvious and objective, things aren’t as clear as we’d like. Every long career will likely include some horrible years, with runs of rejected papers and of grant applications apparently turned down for no good reason. And while it’s sometimes possible to tell the story in a “now we can laugh” kind of way - for example how the paper which would eventually lead to effective COVID vaccines and Nobel Prizes was desk-rejected by Nature, Science and Cell - it’s also worth remembering there were likely other great ideas which never survived the review process, or where fragile postdoc careers ended without the chance to demonstrate how things could have been.

So, if you are at an early stage of your career, it’s worth remembering the role of random chance in how things evolve. Of course, you should do things to the best of your ability, and you try to focus your efforts wisely on projects that can make a difference. But you shouldn’t beat yourself up too much when things don’t work out.

If you have managed to survive this stage, and been promoted into a role where you manage other people, then it would be wise to remember that at least some of your own success can be attributed to luck in the same way. Certainly if you find yourself in a position where you are scoring potential employees to four significant figures, then you should definitely consider that there might be more to predicting success than that, and that the best thing you can provide might be stability to allow people to thrive over long periods.

1. Paperback out 11th April. Time to place a pre-order if you don’t have the hardback!