The European soccer season gets underway this weekend, with three of Europe’s big five leagues — the English Premier League, Spanish LaLiga, and German Bundesliga — all beginning their 2021-22 campaigns. (The French Ligue 1 began last weekend, and Italy‘s own Serie A kicks off on Aug. 21.)
Before last season, I put together a big preview that involved different statistical factors: efficiency, ball control, field factors and regression-to-the-mean factors. The transfer window doesn’t close for a few more weeks, so the rosters aren’t all sorted out just yet, but as we prepare to dive into the marathon that is a European club season, let’s see what some of those stats — particularly that last category — can tell us about the season ahead.
First: let’s talk about regression-to-the-mean factors
We can make pretty good projections based simply on how a team did last year and how much money they’ve spent since last year ended, but some teams will still see the benefit (or liability) of extra bounces in a given season.
Liverpool‘s 2019-20 squad, for instance, was the perfect example of what regression-to-the-mean factors look like. Jurgen Klopp’s squad was clearly awesome overall, but the Reds were also a little too successful in close-game situations — and probably a little lucky in the injuries department — to maintain quite that high a level moving forward. Sure enough, they went from averaging 2.50 points in matches decided by 0-1 goals to averaging a far more mortal 1.43, which cost them about 20 points in the table alone. Factor in a devastating number of injuries (Virgil van Dijk, Joe Gomez, Diogo Jota, Joel Matip etc.) and Liverpool had to scramble just to maintain a spot in the Champions League.
Injuries alone make Liverpool a massive progression-to-the-mean candidate this season, but let’s walk through some pretty common regression factors to see who else we might be under- or overestimating. After all, these factors gave us pretty good hints regarding the regression of not only Liverpool, but also teams like Real Madrid, Juventus, Lazio and Marseille were facing (and the improvement Manchester City was likely to undergo).
So what does it have to say about this season?
Two things to understand about close-game performance: (1) The better teams are better at it. Duh, of course they are. Using a team’s overall points per game, you can run a regression to come up with a pretty accurate projection of how they should be doing in close games (a.k.a. matches decided by 0-1 goals). (2) Some teams are still a little too good or bad in close games in a given year, with Liverpool 2019-20 being a very good example.
Another good example: last year’s PSG. Les Parisiens were good enough overall that they should have expected to average close to two points per game in their close matches. Instead, they averaged 1.29. It cost them about 12 points in a Ligue 1 race that they lost by a single point.
From a close games perspective, PSG (-0.84 points per game compared to expectation) were by far the least lucky team in Europe’s Big 5 leagues. (You could say their luck has turned around in a different way this offseason.)
Compared to what you would expect from their overall performance, others were dragged down a bit by the close-games god as well.
That Inter Milan still comfortably won Serie A while playing merely solid ball in close games (1.9 points per game, same as Milan and behind Lazio) is at least a slight consolation for all the talent they’ve seen walk out the door this offseason — manager Antonio Conte, forward Romelu Lukaku, wingback Achraf Hakimi, etc. Meanwhile, Borussia Dortmund had a strange propensity for giving up early goals — from late-November to late-January, they allowed five goals in the first 15 minutes of matches and pulled just four points from those five matches — that left them with quite a bit of work, and most times too much work, to do.
A few teams also pulled a few more points from these matches than one would have expected. Crystal Palace offered very little last season — 14th in goals scored, 18th in goals allowed — but stayed comfortably out of the drop zone in part because they won seven one-goal matches. They actually averaged more points per game in close games (1.53) than RB Leipzig (1.52), though that probably won’t be sustainable without considerable overall improvement.
Julien Laurens looks into Barcelona’s financial issues leading into their LaLiga opener against Real Sociedad.
One other way to look at how results didn’t actually reflect performance is to look at how well teams played in losses. Five teams averaged an xG differential greater than +0.20 in losses — meaning, their shots in those matches produced an expected goal figure at least 0.2 higher than their victorious opponents on average.
In 2019-20, Manchester City and Inter each finished second in their respective league races, but their average xG differential in losses (+0.47 and +0.24, respectively) suggested they were unlucky to have lost as many matches as they did. That was a portend of things to come.
Surprisingly, Lille led the way in 2020-21, with an xG differential of +0.77 in losses. That can serve as a reminder that they didn’t just win the league because PSG got bad breaks in close games — they were close to putting a higher point total on the board as well. Similarly, Manchester City was second at +0.50. But maybe the most interesting teams to note are RB Leipzig (+0.38) and Sevilla (+0.30). As mentioned above, Leipzig averaged fewer points per close game than Palace did; they’re likely ready for some progression in this regard. Meanwhile, Sevilla finished only nine points back in the La Liga race and probably shouldn’t have lost as many matches as they did.
Saves and set pieces
Save percentages and set piece execution are two things that you have a good amount of control over, but having too much (or too little) success in this category is generally going to be a sign of unsustainability.
StatsPerform has a measure called xGOT — expected goals for shots on target — which is a post-shot measure designed to look at the quality of ball placement for shots on goal. (The commonly used versions of xG are pre-shot measures.) We can use your and your opponent’s xGOT averages (per shots on target) to create an expected save percentage number, and comparing a team’s save percentage margin to this expectation, we can see which teams overperformed in this regard.
Perhaps no one overachieved on set pieces more than Monaco, which scored on 19 of them in league play. That certainly aided their charge to third place in Ligue 1, and while they will likely remain good in this regard, that high a total will be tough to replicate.
I would say the same thing about Atletico Madrid, which dramatically overachieved from the perspective of save percentages — based on opponents’ xGOT per shot on target, Atleti’s save percentage should have been around 71% last season but was 84% instead. But since goalkeeper Jan Oblak is a human save percentage cheat code, this overachievement might remain.
Cards and discipline
Another regression factor that can make a difference in a given match: red cards and second yellows. They can dramatically alter a given match, and if a few more of your matches than normal were altered, good or bad, it can impact your spot on the table.
Easily the most surprising name among the teams with unsustainably high cards last season: Juventus. They suffered six reds and second yellows, and while two those happened late in matches, four happened in the 62nd minute or earlier. They pulled five points from those four matches, a loss of perhaps three or four points compared to expectation.
I’ve been tinkering with game-state information of late — what happens when a match is tied, when a team is ahead/behind, how much time is left, etc. A single goal has a massive impact on results, obviously, and I’ve found it telling to see who might have been a little bit lucky or unlucky, especially when matches were tied.
We often compare goal differential and xG differential to look at whose results might be a little bit off-kilter, and while some teams are designed to over- or underachieve against its xGD, it’s still telling to look at who was the “luckiest” in this regard when matches were tied.
Surprise league champions Lille and Atletico had goal differentials of +1.06 and +1.04 per 90 possessions, respectively, while tied, but those figures were +0.33 and +0.28 higher than their xG differentials in those situations. They were both perhaps a little better than they should have been when it came to building leads, which probably accounts for some of their rise even if you account for the fact that both had great keepers — Atleti’s Oblak and Lille’s Mike Maignan, who has since left for AC Milan. That Tottenham Hotspur had a goal differential +0.32 higher than its xG differential is interesting; Spurs blew a lot of leads last season, and some of those leads might have evidently been undeserved.
One more interesting piece of game state data: who pulled the most points from matches in which they trailed? You would expect the best teams to show up here, and they do, but the top name is interesting. Manchester United (1.82 PPG) was by far the best in this category, but the Red Devils also had to do well in this category since they were more mediocre than other Premier League contenders when matches were tied — while champion Manchester City’s goal differential in tie situations was +1.9 per 90 possessions, United’s was only +0.7.
Regression info aside, let’s look at what else stats can tell us to set the table for the coming season.
English Premier League
(Before we get started: “Since Feb” in these tables refers to teams’ points per game and xG differential from Feb. 1 onward. It’s intended to look at which teams were in good or bad form later in the season.)
Chelsea‘s Champions League run wasn’t a fluke
They needed a last-minute misstep from Leicester City to finish fourth in the league, but you could make a solid case that from the moment they hired Thomas Tuchel to replace Frank Lampard on Jan. 26, Chelsea was the best team in the Premier League. The Blues’ points-per-game average was still inferior to champion City’s from February onward, but xG differential suggests that they were perhaps a bit unlucky not to generate more points (they suffered five draws and two one-goal losses).
Then again, they underachieved their xG figures all year in attack. Expensive new additions Timo Werner and Kai Havertz combined for 18.8 expected goals, but only 10 actual goals; that should progress toward the mean, but how much? For obvious reasons, the addition of Romelu Lukaku will make a massive difference here, but if Werner in particular finds a better rhythm, Chelsea will be a terrifying team.
Is West Ham’s weirdness its own regression factor?
Here are the possession rates for each of last year’s top nine teams:
David Moyes and his lanky Hammers crafted a fun and unique winning recipe out of set-piece glory, quick transition strikes and old-fashioned, hunkered-down, minimal-pressure defense. It was a brilliant underdog approach, it suited the personnel perfectly, and it worked well enough to earn them a spot in the Europa League. But can it work twice? Will opponents figure out ways to adjust, especially with that set-piece goals figure likely to regress at least a little?
Key subplot: transition defense
One of the reasons City underachieved in 2019-20 compared to its baseline stats was its inability to prevent high-quality shots. Within Europe’s Big 5 leagues, only Getafe allowed fewer shots per possession than City, but City also allowed 0.14 xG per shot, fifth-most in the Big 5. Transition defense was a massive issue for them.
In 2020-21, City cut that average to 0.12 xG per shot and put a lot more pressure on shooters. Opponents attempted 71% of their shots under moderate to high pressure, per Stats Perform, which ranked 12th in the league, and only 18% of opponent shots took place with fewer than two City defenders between the shot and the goal. That ranked 18th in the league, but it was more than good enough.
Liverpool’s transition defense, meanwhile, completely fell apart, allowing a league-worst 27% of shots with fewer than two defenders between the shot and goal.
As you can tell — Manchester City were third-worst, Leicester fourth-worst — teams that attempt to play a solid possession game with a pretty high defensive line are going to rank pretty low in this category, but Liverpool’s average was about 50% lower than it should have been. Considering all the injury troubles the Reds had in central defense last season, this makes sense. But let’s track this average all year to see which possession teams are doing their best to prevent high-quality transition chances.
Top five prediction: Manchester City, Chelsea, Manchester United, Liverpool, Leicester City
I’m really going out on a limb there, huh? The only change I’m making compared to last year’s table is bumping Chelsea from fourth to second, but it’s hard to talk yourself into anyone else making a quality run at the top four. Liverpool should rebound a bit simply from the injury bug biting a bit less, and the other three teams were by far the biggest swingers in the transfer window. (And they might not be done swinging yet.) They have been the top four, in some order, for each of the past two seasons, and they appear to have distanced themselves.
The race for fifth, for what it’s worth, could be a doozy. We obviously don’t know what to expect from Spurs until we know whether Harry Kane is still on the roster after the transfer window closes, but Leicester continue to make savvy moves like the addition of attacker Patson Daka, so I’m giving them the benefit of the doubt.
It’s easier than normal to talk yourself out of picking Bayern
The 2020 Champions League winners pulled away for their ninth straight Bundesliga title last season, but they weren’t quite as convincing as they had been after hiring Hansi Flick late in 2019, and they’ve had a few shaky transfer windows in a row.
In this window, they have spent most of their transfer budget bringing in manager Julian Nagelsmann — Flick left to become manager of Germany‘s national team — and centre-back Dayot Upamecano from RB Leipzig. Those are quality additions, to be sure, but they will still be relying heavily on one-year-older attackers Robert Lewandowski and Thomas Muller, and both winger Kingsley Coman and midfielder Leon Goretzka are stuck in contract disputes, and right-back, already a position of minimal depth, recently took a hit with Benjamin Pavard‘s ankle injury.
If another contender has its act fully together, this could be a prime opportunity for Bayern’s nearly decade-long title streak to end.
Sebastian Salazar and Herculez Gomez discuss Josh Sargent’s move from Werder Bremen to Norwich.
Borussia Dortmund lost Jadon Sancho to Manchester United, but they brought in PSV’s Donyell Malen to compensate, and plenty of key youngsters like Jude Bellingham and Gio Reyna (not to mention 16-year old Youssoufa Moukoko) will likely improve. Oh, and they will evidently still have all-world goal-scorer Erling Haaland for one more year. If they get a few more breaks in tie-game situations and aren’t chasing the scoreboard as much, they could at least make a run at 75 points or so this year after hitting only 64 in 2020-21.
RB Leipzig might be the more interesting contender, however. The 2020 Champions League semifinalists have one of the most exciting collections of attacking talent on the planet, and while there’s no guarantee that the defense will hold up, American manager Jesse Marsch’s first Leipzig squad will have quite a bit to offer in resistance to big, bad Bayern.
Once again, transition defense will be huge
Opponents found themselves far more capable of hitting Bayern on the counter last year; that was a big reason for their loss to PSG in the Champions League quarterfinals, and it was an issue in league play as well.
As with Man City in 2019-20, you didn’t get many shot attempts against Bayern, but the chances you got were very strong. They ranked eighth in xG per shot allowed, and a general lack of shot pressure allowed opponents to place those shots really well:
Only 65% of opponents’ shots against Bayern were under moderate to high pressure, second-fewest in the league behind Leverkusen.
Nagelsmann’s RBL were better in these categories, and we should probably expect a bit more of a pragmatic approach this year, but how much? And what happens if the pragmatism also tamps down Bayern’s attacking upside?
Top five prediction: Bayern Munich, RB Leipzig, Borussia Dortmund, Borussia Mönchengladbach, Bayer Leverkusen
I can’t do it. I can’t pick against a nine-year streak. I expect RBL and BVB to both improve on last year’s figures, and Bayern could legitimately be in a serious dogfight here, but they still probably have the most to offer overall.
Meanwhile, though I’m very curious about what Eintracht Frankfurt might be capable of in attack, my guess is that Gladbach, which was a bit unfortunate not to finish higher last season (and which stole manager Adi Hutter away from Eintracht when Marco Rose left for Dortmund), might have the inside track to the No. 4 spot thanks in part to the fact that they don’t have to worry about a continental competition this time around.
What does Messi’s departure actually mean?
Despite all the drama last year — Leo Messi asking to leave, Luis Suarez getting sent to Atletico, etc. — Barcelona was still the best LaLiga team on paper last season. They scored by far the most goals, allowed the fourth-fewest, and lapped the field from an xG perspective. They remain third in 538’s club ratings for now, too. None of this added up to a title, obviously — they dropped from 82 points to 79, seven points back of Atleti. But if the club had made no changes whatsoever, they would have probably been considered the league favorite heading into 2021-22.
Instead, the club is undergoing its biggest facelift in nearly 20 years, and not only because the long-term face of the franchise is gone.
Gab Marcotti questions Barcelona’s decision-making when it comes to failing to come to agreeable terms with Lionel Messi to remain with the club.
With Barcelona finding out that money is actually real and they couldn’t navigate around LaLiga’s entirely sensible salary restrictions in a time of massive debt, Messi was forced to leave for PSG, and Barcelona was forced to rely primarily on free transfers — albeit high-quality ones like Lyon‘s Memphis Depay, Manchester City’s Eric Garcia and the already-injured Sergio Aguero. We’ll see what other moves they can make (and whether they can get any further acquisitions registered to play), but Barcelona is going to look awfully different this year.
Of course, from a raw talent perspective, they’re still going to have more of it than anyone but maybe Real Madrid. They’ve still got Antoine Griezmann, young stars Frenkie de Jong, Pedri and Ansu Fati. Acquiring players like Griezmann and Philippe Coutinho may have destroyed the club’s finances, but they’re still talented. If Ronald Koeman can stitch together something cohesive, Barca could still be a La Liga contender. We just don’t know if he will, or how long it will take.
If there’s a leadership void, who fills it?
Again, odds are solid that Barcelona and/or an aging but still openly talented Real Madrid will still loom atop the La Liga table despite their ongoing financial difficulties and minimal transfer opportunities. And Atletico Madrid returns to defend its title with most of its squad intact, even if said squad faded a bit and barely clung to its lead late in the season.
This is still the best opportunity a club outside of Spain‘s Big Three has had in quite a while to make a move, however. Who’s most likely to do it?
Sevilla obviously has a solid chance. Julen Lopetegui’s squad stretches you from side to side and defends like crazy, and as you see in the table above, they were every bit the contender on paper as they were in the win column, finishing just two points back of Barca.
Real Sociedad started and finished well but fell victim to two lengthy slumps (nine points in 11 matches in December-January, two points in five matches in March-April) and finished far back of the field. But in terms of xG differential, they topped both Sevilla and Real Madrid. They created far cleaner looks than their opponents — 34% of their shots came under light or no pressure (fifth in the league), and only 25% of opponents’ did (first) — and if they finish their chances better, they could make a move.
Villarreal couldn’t make a sustained push in league play, but they showed what they could do in winning the Europa League, and Unai Emery now enters his second season in charge.
Don’t write off Real Betis, a team that improved quite a bit down the stretch and returns most of last season’s squad.
Top five prediction: Atletico Madrid, Real Madrid, Sevilla, Barcelona, Real Sociedad
I am quite obviously not a bold risk-taker when it comes to predictions, but while Atleti could be in trouble if Oblak is ever less than 100% amazing, it’s hard to make the case that either Real Madrid or Barca will top even last year’s numbers, while the team that is most likely to improve in my eyes, Real Sociedad, is starting from so far back.
We aren’t that far from a nutty title race, but the standard favorites remain the favorites.
Italian Serie A
Atalanta might never have a better shot
As a club, Atalanta is at a high ebb. They made the Champions League quarterfinals in 2019-20 and the round of 16 last year — they were eliminated by Real Madrid after getting hamstrung by an early red card at home — and they’ve finished third in Serie A for three straight seasons.
The results have been brilliant for a club of smaller stature, but the what-ifs have been clear too. For two straight seasons, they have labored through league play while focusing on the Champions League group stage, only to surge when said group stage ended.
That they finished third again in a tight battle is an incredible accomplishment, but especially after falling just five points short of the Scudetto in 2019-20 following a similar surge, there has to be a lingering hope for something greater.
With the core of Inter’s title team getting torn apart and second-place AC Milan spending most of its transfer budget to date on a replacement for goalkeeper Gianluigi Donnarumma (Lille’s Maignan) and the permanent transfer of last year’s Chelsea loanee Fikayo Tomori, this might be Atalanta’s best shot.
They’re dealing with their own turnover, as always, but they’re pretty used to that, and they’ve made their own defensive upgrades in bringing in players like Udinese keeper Juan Musso and defenders Matteo Lovato (Hellas Verona) and Merih Demiral (Juventus).
After seeing its long title streak snapped last season, Juve starts the season the betting favorite once again. After last year’s poor fortune in close games, that might make sense even if we don’t know how they will finish up in this transfer window. But if Atalanta can ever reach December without having to make up huge ground, they could make this an interesting race.
Julien Laurens and Don Hutchison debate whether Inter Milan’s Lautaro Martinez would be good fit for Tottenham.
What about Napoli?
Despite playing maybe the prettiest possession game in the league, Napoli finished a point off of the pace for a Champions League spot. Their 56% possession rate was third in the league, they attempted far more passes and shots than their opponents, and they had both the second-most carries and by far the best ground duel win percentage. They did most things well, but they allowed opponents just a few too many decent looks and didn’t stop enough of them.
Napoli added veteran manager Luciano Spalletti, most recently of Inter, and permanently brought winger Matteo Politano from Inter after a loan in 2020-21. Their squad remains mostly intact, and Spalletti’s possession-friendly ways seem like a decent fit. After altering between fearsome and flaky, the Azzurri might be a wildcard in the Scudetto hunt.
Top five prediction: Juventus, Atalanta, Napoli, AC Milan, Inter Milan
It’s hard to completely write Inter off — they will still have Lautaro Martinez, after all — and Roma is a mystery with Jose Mourinho taking over. But Juve, Atalanta and Napoli appear most well-positioned to fill the void if Inter indeed regresses quite a bit.
French Ligue 1
If PSG hadn’t made a single offseason move, they’d have been likely to rebound and comfortably win the league comfortably
Lille’s title defense was always likely to fail if only because PSG was likely going to perform better in close matches, and Lille might have found fewer breaks in building leads.
But while Lille lost manager Christophe Galtier to Nice and sold Maignan, midfielder Boubakary Soumare (Leicester) and Luiz Araujo (Atlanta United), PSG was piecing together maybe the greatest transfer window performance of all time.
They added the brilliant Achraf Hakimi from Inter and Danilo Pereira from Porto and, ho-hum, brought in Messi, Donnarumma, Sergio Ramos and Georginio Wijnaldum via free transfers. PSG was strangely short of matchup advantages beyond Neymar and Kylian Mbappe, and now they have nearly an entire lineup’s worth of them. It’s hard to pretend that anyone in Ligue 1 will be able to keep up with this all-star team for too long into the season.
Craig Burley explains why clubs should pay whatever they have to in order to land Dortmund’s Erling Haaland.
That said, the race for second place could be fascinating
You’ll probably end up watching a bit more of Ligue 1 this year if only to see the incredible things a Messi-Mbappe-Neymar attack can conjure up. But while you’re there, watch the other teams, too. FiveThirtyEight gives five teams between a 15-46% chance of earning a Champions League bid this year, and it’s not hard to make the case for quite a few of them finishing second.
Monaco was equal to both PSG and Lille down the stretch and is brilliant in the set-piece department (even if they probably can’t count on 19 set-piece goals again). They possess the ball well and allowed the fewest shots per possession last year. Predictably, they also had a bit of a transition defense problem, but they were awesome late in 2020-21.
Lyon was first in the league in xG differential and third in goal differential. They were the best possession team besides PSG, but their defense was just leaky enough to allow Monaco to pass them for third. Manager Rudi Garcia and forward Memphis Depay both left, but there’s still talent here.
Marseille has been excitingly competent since bringing in both a new president (Pablo Longoria) and manager (Jorge Sampaoli) following a fan riot last February. They are loading up on high-upside youngsters like Arsenal’s Matteo Guendouzi (loan) and Barcelona’s Konrad de la Fuente (Barcelona), and while they could be a flighty and streaky team — as evidenced by their first match of the season, in which they fell behind 2-0 then surged to win 3-2 — they’re going to be far more entertaining than usual.
Lille is replacing key defensive pieces, and new manager Jocelyn Gourvennec has never managed a club with high expectations before, but they still have midfielder Renato Sanches, attackers Burak Yilmaz, Jonathan David, Jonathan Bamba and Tim Weah, etc. Raw talent will give them a shot, though you have to worry about the drain on depth Champions League play could have.
Rennes was sidetracked by one simple issue last year: they couldn’t shoot. They controlled the ball (59% possession rate) and shot far more than their opponent. And they still have midfielder Eduardo Camavinga (for now) and some of the other reasons for this ball quality. But you have to put the ball in the net, too.
Top five prediction: PSG, Monaco, Lyon, Rennes, Marseille
It’s had to know what Lille is capable of, especially with the core of a defense-friendly system getting torn apart, but the second- and sixth-place teams might not be particularly far apart this season.