Is the 'new manager bounce' really all a myth?

I'm just asking questions here...

With Mauricio Pochettino sacked and Jose Mourinho on to replace him, Tottenham beat West Ham 3-2 on the weekend. I thought, therefore, this might be a good time to reassess the idea of the ‘new manager bounce.’

Is it real? Or an illusion?

The last word (or the latest word) on the new manager bounce seems to come from Simon Kuper and Stefan Szymanski’s Soccernomics, which definitively put the notion to bed.

Here is Kuper, for example, writing about it in FourFourTwo back in 2016:

In fact, new managers cannot work magic. The short honeymoon after the new man takes over is easy to explain. Typically, the average club earns 1.3 points a match. Typically, Bridgewater found, a club sacks its manager when it averages only 1 point a match—that is, at a low point in the cycle.  

Any statistician can predict what should happen after a low point: whether or not the club sacks its manager, or changes its brand of teacakes, its performance will probably “regress to the mean” – or in ordinary language, return to normal. Simply put, from a low point you are always likely to improve. The club may have hit the low due to bad luck, or injuries, or a tough run of fixtures, or—as perhaps in Manchester City’s case in 2009—the time it takes for a largely new team to gel. Whatever the reason for hitting a low, things will almost inevitably improve afterward.

Google ‘new manager bounce’, in fact. and you will be treated to variations on this theme. Here’s Pinnacle:

Analysing managerial turnover across 18 seasons (1986 to 2004) in the Eredivisie, Bas Ter Weel revealed noticeable patterns of prior decline and subsequent improvement centred on the sacking of one manager and the appointment of a new one. Crucially, however, almost the same pattern could be observed where managers had not been sacked. How so? Ter Weel was unequivocal in his explanation:

“If managers do not matter for differences in performance... and quality does not vary across managers, the only observed performance change following turnover would be mean reversion.”

And the Independent:

The ‘managerial bounce’ is a chimera: this phantom bounce is nothing more than regression to the mean. On average, an in-season replacement of the manager has zero effect on performances, either during the season or in the long run.

And here is friend-of-the-newsletter Omar Chaudhuri quoted by Reuters:

“What we see is that when managers are sacked, a lot of the times their teams have been playing okay, but without luck. It only needs one crucial moment per game to go against you and that is the difference between three points and one or one point and no points. In most cases sacked managers are unlucky.

“It’s like calling tails five times in a row in a coin toss and losing each one. The next person calls tails and wins. It’s the same with a football manager.”

Now, I should stress that while I do not disagree with this thesis, I have several quibbles with it.

What do we mean by regression, exactly (get it)?

First, these arguments seem to cloud what regression to the mean actually is.

To demonstrate, let’s say you flip a coin and get six heads in a row. “Damn,” you say to me, who’s standing next to you for some reason: “Maybe you should try it to see if the coin isn’t weighted.” So then I flip it and get four tails in a row. Voila! It’s the new coin-flipper bounce! Which we all know is actually just regression to the mean; if we both spent all day going back and forth flipping coins, our results would move closer and closer to 50/50 heads and tails: it’s all random!

But that randomness means it’s possible that when you hand me the coin, I might get something like three more heads, one more tail, and three more heads again. Maybe we figure the coin stinks and we’ll need to get a new one in the January transfer window. But of course, it’s still all just random and has nothing to do with the flipper.

In the same way, if a manager is just a very well paid and overhyped coin-flipper, hiring a new one shouldn’t ‘trigger’ regression to the mean. While some might get that sweet bounce, other managers might ‘take some time to right the ship’ (lose for a while and then win a few), or ‘get uneven results’ (win, lose, win, lose). I think we do see this, in fact (but I would like someone who is smarter than me to show me the data on this)!

What I’m saying here is regression or reversion to the mean shouldn’t be triggered by a new manager. Switching the flipper isn’t a guarantee you won’t keep getting mostly heads for a while. So this, to me, can’t explain the common perception that there is a bounce.

Second, I don’t think there is anything about the colloquial understanding of the ‘new manager bounce’ that necessarily implies it’s either caused by the new manager or is the start of permanent long-term change. It’s literally called a bounce, as in up briefly and then down again. The chimeric quality of the new manager bounce is built into the very concept, so we need to be careful about whose myth we’re actually busting here.

Even before Soccernomics, the majority of fans, having been around the block before, were not under any illusions that hiring a new manager would immediately meaningfully elevate a team’s long-term fortunes on its own. Familiar comments like “let’s see what moves they make in the next window” or “they’ll need time to build the squad before we can properly judge them” tend to underscore that fact.

That said, there is still a sense from your casual fan/pundit that there is a bounce, and that it happens relatively frequently. I’m not going to say they’re correct as a rule, but I’m also not going to say the perception that teams play differently after hiring a new manager is entirely fantastical, either.

Third, not every managerial sacking is the result of bad luck (ie a string of heads in a row), which slightly undermines the mean regression thesis. I would say that historically most managers have been sacked because of poor luck; clubs didn’t have the advantage of metrics like expected goals until recently, which would have helped offer an anchor for overall performance against table points.

In Spurs’ case, however, the team’s results were more or less precisely on par with their expected goals for and against in the six matches leading up to Pochettino’s departure, according to understat.com (other models may differ). So Poch wasn’t sacked for bad luck. He was sacked because the team’s underlying performance was poor. Spurs are bad!

Yesterday, in Mourinho’s first game in charge, they beat an also bad West Ham side 2-3, though it was an away performance—their first away win since January—in which Spurs scored 1/7th of their season goals total so far. We were even treated to Zonal Marking’s tactical analysis of new-look Spurs after a single game after Mourinho was in charge for less than a week! Miracle worker indeed!

So is the bounce real?

Joking aside, while I think the new manager bounce, as a rule, is largely a myth, I don’t think the emotionally charged process of sacking a long-time manager and replacing them with a new one has no effect on performance whatsoever, and I don’t think people are dim for thinking so.

First, players are usually among the first to get a sense of where the winds are blowing ahead of a sacking. As I’ve argued recently in this newsletter, they’re only human, and a team isn’t going to play the same way for a dead-man-walking—an underfire manager they may consciously or unconsciously blame for their current struggles—than they will play for someone who represents a fresh start and a break from the past. I doubt this ‘effect’ is large enough to rock the xG scales in any meaningful way, but I don’t think it’s entirely non-existent and non-perceptible to your everyday football supporter.

Second, though I am on the record as (facetiously) questioning whether clubs really need managers at all, to say that they have no effect on team performance, even over a handful of games, seems a stretch. Of course, there are enough bad managers who coast on recognition alone who would skew any meta-analysis on the new manager bounce, but sometimes clubs do indeed sack good managers and hire great ones (see Southampton in the early Les Reed days). It’s also theoretically possible that a new manager might come in, change tactics, and do slightly better for a while until opposition clubs catch up to what they’re up to and adjust accordingly.

All this is to say I think simply looking at large, multi-team studies that clearly show regression to say there is no such thing as a new manager bounce full stop is, perhaps, a little naive. Ultimately, however, the new manager bounce phenomenon is pointless, a blip, a curiosity. Teams should hire managers for the long-term, with a clear plan in place for how they will improve and sustain performances relative to the quality of the squad. The jury’s out whether Mourinho can do with until Luis Campos decides if he wants to join the party, but for now—bounce baby bounce.