FluxVlog

This has been the season of sackings in the Premier League.

Let’s start with the basic facts. The 2022-23 campaign saw the most sackings ever in a Premier League season, with 14, from Scott Parker in August to Javi Gracia in May. The previous record was 10. There were more sackings this season than in the first seven Premier League seasons combined.

Advertisement

In all, 11 clubs changed manager, which is the first time that more clubs made a switch than didn’t. Of those, Brighton were the only club forced into a change, when Graham Potter left for Chelsea.

It’s the first season that more than one club made more than one change: four did in fact, with Chelsea, Southampton, Tottenham and Leeds all dipping into the big manager tombola multiple times.

Elsewhere on The Athletic today…

This season saw one of the latest sackings by a team still with something to play for in Premier League history, with Leeds deciding that Gracia could not be trusted with their final four games and rolling the dice with Sam Allardyce.

In total, if you include caretakers, 39 men took charge of at least one Premier League game.

The reasons why there have been so many sackings are varied and complicated, so we won’t get into that here. Instead, let’s take a look at which ones actually worked…

Firstly, the most basic metric: league positions.

Six clubs only sacked one manager and all of them finished in a higher league position than they were when the trigger was initially pulled. One of those is Leicester, who moved from 19th when Brendan Rodgers was dismissed to 18th by the final day. Which, as scholars of the league table will tell you, is not much use. But the rest experienced more tangible improvement.

Crystal Palace only actually moved up a single place and Bournemouth three, but that doesn’t really tell the story of their progression under Roy Hodgson and Gary O’Neil respectively. We’ll get to that shortly.

Everton went from 19th to 17th: only two places, but of course they crossed that big red line and survived. Only just, in the end, but the horrors that would have come with relegation mean they were justified in replacing Frank Lampard with Sean Dyche.

Advertisement

Wolves moved up five places after Bruno Lage was dispensed with, but as you will probably have guessed by now, the change to have the most significant impact on a club’s league position was Aston Villa shipping out Steven Gerrard and shipping in Unai Emery. They went from 17th and the brink of the relegation zone to seventh and European football next season.

Things didn’t turn out so well for the four clubs that got rid of more than one manager. Chelsea were sixth when Thomas Tuchel departed, had slipped to 11th by the time they lost patience with Graham Potter and ended the season in 12th, their lowest league finish since 1993-94 and on 44 points, their lowest total since 1988, when they were relegated.

Southampton were 18th when Ralph Hasenhuttl departed, had slipped to dead last under Nathan Jones and they stayed there under Ruben Selles. Tottenham were fourth when Antonio Conte finally went, had dropped one place after that Newcastle game which saw the end of Cristian Stellini and finished in 8th.

And then there’s Leeds. Oh, Leeds. Jesse Marsch had them just above the bottom three in 17th when he was fired, they were in the same spot after Gracia’s brief spell but dropped to 19th, and out of the division, after four games of Sam Allardyce.

League positions only tell some of the story, though, as the O’Neil and Hodgson cases show.

What about points per game? Well, again it’s an imperfect metric, but it’s still worth looking at a few cases.

Again, Aston Villa can feel pretty good about their decision to bring in Emery over Gerrard. Under the former England midfielder, they gained 0.82 points per game but with the Spaniard that jumped to 1.96.

Almost as big a leap came at Crystal Palace. With Vieira in charge, they managed a neat 27 points from 27 games but, after the freewheelin’ Hodgson’s return, that leapt to 1.8 PPG. Stretched out over a season, that would have been enough for 68 points and fifth place, above Roy’s old friends Liverpool.

Advertisement

For O’Neil, these numbers require a little finessing. Without context, Bournemouth got 0.75 PPG under Parker and 1.06 after he left in August. However, the sample size under Parker (just four games), is too small to take much notice of. Plus, while from a statistical point of view we probably shouldn’t do things like this, if you lop off the Cherries’ last four games after survival had been assured and the players were rubbing suncream on each other’s backs and building sandcastles, O’Neil’s PPG jumps to a slightly more flattering 1.2.

When Julen Lopetegui is making his case for more backing to the Wolves board, he could do worse than throwing his numbers up onto the big screen: his 23 games saw them gather 1.34 PPG but, in the 15 fixtures under Bruno Lage and caretaker Steve Davis, that figure was only 0.59.

Southampton’s numbers tell us pretty much what we already know: they were bad under Hasenhuttl (0.86 points per game), astonishingly bad under Jones (0.38) and went back up to merely very bad under Selles (0.63). All that said, anyone doing PR for Hasenhuttl might want to point out that if they had continued to pick up points as they had under him, they would have been on around 33 for the season: not enough for survival on its own, but they would at least have just about been alive for the whole season, rather than adrift from around March.

Read on face value, the points per game tell us that Tottenham should have stuck with Antonio Conte: they got 1.75 under him, then 1.1 with Cristian Stellini and Ryan Mason. However, the numbers don’t factor in things like “astonishingly toxic vibes”, so clearly more context is required there.

There is a more serious case to be made that Leeds might have been a bit hasty when canning Marsch. They picked up 0.9 PPG across 20 games under the American, whereas in the other 18 under three different managers, they dropped to 0.67. And as Gracia has pointed out, he gathered 11 from 11 games, so you could say it was the other seven, under Michael Skubala and Sam Allardyce when they collected a miserable two points, that relegated them.

Points and league positions only tell some of the story, though. How about a nice big slice of expected goals?

Looking at the data, we can use rolling expected goals (xG) charts to track how teams have been performing throughout the season. The blue lines represent a 10-game rolling average of xG for — an indication of how many goals a team are expected to score based on the quality of chances they create — while the red lines chart xG against, allowing us to see how many goals they are expected to concede.

Advertisement

By averaging this data across the previous ten games, it provides us with an indication of how a team have generally been performing at a given point — a higher red line means that the team are conceding more chances than they create, while a higher blue line tells us that their attacking output outweighs the quality of opportunities that they allow.

There are a few pretty interesting things to draw from these charts, the first being: should Chelsea have given Potter more time?

As this chart indicates, things didn’t start off brilliantly after his move from Brighton, but you could argue it was just the continuation of a trend from Tuchel’s last days. That trend lasted longer than it should have done, but towards the end of his reign the numbers seemed to look a little more favourable, only for that trend to immediately nosedive once Potter had been replaced by Lampard. There were many other factors in their decision to get rid of Potter but did they give up on the experiment at just the wrong time?

This chart of Leeds United’s season indicates why they might have so quickly pulled the plug on Javi Gracia, the sharp incline in their xG against under his leadership resembling the sort of gradient that only mountain goats can climb.

But there’s more indication that Marsch may have been harshly treated: while you would struggle to describe Leeds, based on these numbers, as ‘good’ under him, neither were they especially bad. And you don’t necessarily need to be ‘good’ to avoid relegation. This is all in hindsight, but might they have avoided their grim fate if they had stuck with Marsch? Or maybe the mistake was not getting rid of him sooner, giving his replacement more time and choosing that replacement more carefully.

Everton’s graph is slightly surprising, in that it suggests Sean Dyche actually made their defence worse, rather than better, and it was an improvement in attacking production that kept their heads above water. Who knew?

Southampton’s chart very slightly raises an eyebrow. It’s no surprise that, regardless of who their manager was, they could barely create a chance all season, but the graph does indicate that arguably their best spell of the season was under… Nathan Jones. Of course ‘best’ is a relative term, but it was under him that their xG against reached its lowest point of the season. Should Southampton have kept patience w… no, actually, let’s stop there, it’s too silly.

There is some nuance at play here, but Leicester’s graph plays into the idea of timing: you can see the point, a few games before they actually did dismiss Brendan Rodgers, when their xG against was touching two and xG for wasn’t too far above one. It’s difficult to chastise a club for not making a move based purely on a rolling xG chart like this, but some hints of improvement under Dean Smith suggest that if they had done so just a few games earlier… they might they have survived.

Here’s a fairly straightforward one, on the face of things: we could probably remove the label that showed where Hodgson took over Palace and you would be able to tell, although this should be considered with a significant caveat in mind.

While the data does show a pretty dramatic improvement in both attacking and defensive performance, the identity of their opponents has to be taken into account: Vieira was dismissed just after a ruinous post-World Cup run which saw them play 12 out of 14 league fixtures (Manchester United and Brighton twice) against teams that finished in the top ten. Their next three games were against the relegated trio, and in their last 10 they only faced two teams who finished in the top half, and didn’t beat either of them. Vieira probably wouldn’t use this chart if asked to prepare a presentation in his defence, but he could include the fixture list on his PowerPoint.

Advertisement

This is less a graph of how a football team has performed, more a chart displaying the gradual decay of an institution. Things started off OK under Antonio Conte, continuing the improvement he inspired in the previous season, but once the rot set in neither Cristian Stellini nor Ryan Mason could do anything about it.

Now here’s an interesting graph. There was some clear improvement after Lopetegui replaced Lage, but as the season went on these numbers suggest performances at both ends of the pitch gradually got worse and worse, culminating in 6-0 and 5-0 defeats to Brighton and Arsenal in the closing weeks.

We should make some allowances for the fact that, from around April, it was fairly obvious that Wolves were not in much relegation trouble so their motivation may have dropped off a little, but while this chart wouldn’t suggest they were wrong to make the change when they did, it might give them pause for thought about what comes next.

Finally, Aston Villa’s metrics are interesting in terms of what they don’t tell us, rather than what they do. Their actual performance hasn’t really tallied with how the data expects them to play, which could mean a few things. No model is perfect, so they could be one example that falls through the cracks of how we measure these things; they could be exceptionally efficient and clinical in attack, thus outperforming their metrics; they could have enjoyed some luck under Emery; their performances are potentially unsustainable and something of a crash could be expected next season.

There are a couple of problems with the question ‘who was right to sack their manager?’, the first being that it’s only half the question: you may have been right to sack your manager, but more important is who you replaced them with.

Another is that there is no counterfactual. We don’t know what would have happened had any of these managers simply stayed in place. We can take a good guess at a few of them, depending on how much stock you put in the data, but we can’t know for sure.

Take Liverpool. Their 3-0 defeat to Wolves on February 4 — which came hot on the heels of a 3-1 loss to Brentford, a 3-0 beating at Brighton and, perhaps most embarrassingly of all, a 0-0 draw with Chelsea — left them in tenth place, 21 points behind leaders Arsenal and closer to the bottom of the table than the top. Many other big clubs who had title aspirations might have sacked their manager in that situation.

They didn’t, because their manager is Jurgen Klopp, and it would have been preposterous. And as it turns out that was the right decision, because after that date Liverpool lost just two of their remaining 18 games, and only Manchester City took more points in that time.

Advertisement

This isn’t to suggest that Lampard would have turned things around had he been given more time at Everton, or Jones was hard done by when Southampton cut their losses, but drawing firm conclusions on who was ‘right’ to make a change is difficult because we will never know what the alternative was.

It’s also not an argument that everybody should have stood pat. It’s been suggested in various places that the sole reason teams like Nottingham Forest and West Ham stayed up was because they stuck with their managers and because the three relegated teams did make changes and that’s why they went down. But that feels pretty simplistic, and there are circumstances particular to every club that informed their decisions.

Even the fact that the three clubs that went down went through ten managers between them, which feels extremely sub-optimal, brings with it queries and nuance: were they bad because they were chaotic, or were they chaotic because they were bad?

This feeds into the argument that the identity of the man in the dugout doesn’t actually matter as much as people think, as outlined here. It’s usually just the teams that have spent the most money, or have the best players, that succeed.

All that said, could you really argue that Bournemouth’s squad, on an individual level, is better than Leicester’s? Brentford spent much less money on their squad than Leeds, so surely the fact that they enjoyed some stability in the dugout is a factor in their success.

Either way, the evidence from this season — and previous ones, as Duncan Alexander explained — suggests that if you’re going to sack your manager, do it early. The biggest successes — O’Neil, Emery, Lopetegui — tended to be the ones done before the World Cup break.

The exception to the rule was Hodgson at Palace, but otherwise every managerial change that occurred after January made very little difference, and in most cases performances got worse, not better.

Advertisement

So, who was right to sack their managers? Aston Villa; absolutely positively, yes. Wolves and Everton; definitely. Tottenham; probably, but don’t let it reach the point where everyone hates each other. Southampton; possibly, but maybe think more carefully about the replacement. Leicester; probably, but they shouldn’t have left it so late.

But Leeds? Chelsea? There are a few managers out there who will think themselves quite hard done by.

GO DEEPER

How Leicester's 5,000-1 Premier League fairytale turned into a nightmare

(Photos: Getty Images)

ncG1vNJzZmismJqutbTLnquim16YvK57k25tbGllbXxzfJFsZmltX2h%2BcLzRnqSinaJiuaatxq6cZqWRo66osdGsZKyZk6CypXs%3D