tactical-substitution

[tactical-substitution]

City vs. xG: A Shot-by-Shot Postmortem of an Attacking Performance

In my first post on this blog I considered the question of whether Liverpool’s dominant run in the league prior to the break could be attributed to luck or good fortune. Part of that analysis involved looking at their primary rival for the title, insofar as they had one, Man City - through the season City have looked like a possession-dominant side capable of piling on absurd numbers of shots, yet their points total is comparatively mediocre. I mostly attributed this discrepancy to their defensive performance, and I still believe this to be the primary culprit in the number of disappointing results that Pep Guardiola’s side have suffered this season. City have not underperformed their xG hugely or failed to breach opposition defenses apart from a handful of games, whereas their defense has let them down at nearly every turn. This has been blamed on the lengthy absence of center-back Aymeric Laporte, though perhaps an even bigger factor has been defensive midfielder Rodri’s period of adjustment to the task of replacing Fernandinho, and the resulting susceptibility to fast attacks which the Spanish midfielder has been unable to slow down or stop. Conceding two goals from two shots against Spurs in the second match of the season has not helped, nor has a late screamer scored by Jonjo Shelvey to equalize for Newcastle United. Whether it has been misfortune or ineptitude which has led to City’s defensive decline, it has been this defensive weakness which principally rendered them unable to mount any kind of serious title challenge after two fairly imperious league title wins.

Since the resumption, Liverpool have sealed the title, City got some measure of revenge by taking the champions apart 4-0 at the Etihad, and it seemed that there was nothing left to write about concerning England’s two current best club sides. And then, mystifyingly, City lost 1-0 to Southampton - in and of itself this is not that vexing a result, the Saints are a reasonably good side and neither club was playing for all that much. However, given that City almost entirely dominated the ball and took twenty-five shots to their opponents’ seven, many of them from good areas, their nightmare season has provided us with yet another opportunity to think about the how and why of football. Fivethrityeight’s xG model has the match at 3.6-0.7 in City’s favour - Southampton good value for their single goal, but how on Earth did City fail to get on the scoresheet? It is exactly this that I will look at, although there truly is no answer beyond “sometimes they just don’t go in”. Variance intuitively seems like a bad explanation for a match of this nature to pattern-seeking cognition, and most people would prefer to think that Southampton defended City’s twenty-five shots in some special way, or that the finishing techniques of City’s forwards was particularly ineffective but, well, when one actually looks at each of City’s shots there is no good reason to think that this is the case. Southampton’s keeper and back-four are alright and looked competent and relatively error-free in this match where they were constantly in the thick of it, but they hardly nullified City’s danger-men. As for City? They were fine, and while the scoresheet doesn’t reflect it this was another attacking bonanza of the type that City have regularly orchestrated over the past three seasons.

So then let’s look at City’s attacking performance through a description of each shot that they took. None of these were tap-ins, some were truly optimistic, but it’s somewhat remarkable that not one found the back of the net.

Minute 11: A “speculative effort” from a free-kick near the halfway line, apparently an attempt to take advantage of Saints keeper Alex McCarthy “off his line”. The commentator does not even mention who the taker was. Not the most auspicious beginning to City’s attacking performance - probably fair enough that one didn’t go in!

Minute 16: Southampton 1-0 City! Not a City shot, obviously, but it does determine the dynamic of the rest of the match. Che Adams catches Ederson “off his line” but it’s not necessarily an outright error - Ederson is positioned correctly to facilitate City’s buildup, and if the error is anyone’s it’s Zinchenko’s, giving away the ball in midfield under pressure. This feels like the goal that Southampton manager Ralph Hassnehuttl constructed his side to score, winning the ball in a seemingly innocuous midfield area with his forward players in position to attack quickly. City should have been anticipating that the Saints would press in this manner, but eh.

Minute 23: City have had the pressure on since they went behind. Joao Cancelo makes space for himself with a quick change of direction in the box and cuts it back for Gabriel Jesus at the far post, who is unable to put it away. It’s an alright chance but the cutback is played with a little too much pace for Jesus to finish it with any precision.

Minute 27: A nice little move starts at the feet of Laporte, leading to a corner. City don’t play it right into the box but rather work it to Cancelo, who plays a ball over the top for Sterling, who runs in behind and plays a cutback for Jesus. It doesn’t quite connect and Danny Ings clears it, but it’s good play that could well have led to a goal.

Minute 30: Plenty of excitement here. Cancelo, who’s been having a good match, feeds the ball into Jesus’s path in the right channel, and Jesus finds Sterling in the box. Sterling shoots powerfully with his left foot but McCarthy saves; City collect the rebound and Fernandinho shoots from outside the box, hitting the post. Bernardo Silva floats a cross for David Silva, who heads it straight at McCarthy, and Sterling thwacks it high and wide. "How is it still one-nil?” asks the commentator.

Minute 32: David Silva just undercooks a through-ball for Jesus to run onto, which could have been a very good chance.

Minute 37: Cancelo wins it off Nathan Redmond in midfield and carries it to the edge of the box - he’s been excellent. He slips in Mahrez, who shoots from a wide angle. McCarthy saves with an outstretched leg, it ends up back at the feet of Mahrez, who tries another shot which is blocked by Bednarek. Given the Algerian’s ability to curl one in with his left foot these were definitely decent chances.

Minute 45+2: Some hypnotic buildup play in which every City player is involved gets Bernardo Silva into a position on the right side of the box. He dinks a cross for Jesus to try and head, but the relatively diminutive Brazilian loses his header.

Minute 48: City win a corner. Mahrez takes, Jesus gets a glancing header that just goes wide. Decent enough chance.

Minute 50: It is, as you would expect, all City. Sterling gets into a pretty menacing position taking on Stephens against the by-line in the box, Stephens pokes it away for a corner. City circulate the ball to the right side of the pitch where Fernandinho delivers it into the box. Laporte shoots from a central area but it’s saved; Jesus tries to put away the rebound but McCarthy saves that as well, from point blank range.

Minute 51: Zincehnko shoots from distance, Ryan Bertrand blocks it. That technically counts as a shot, though it’s hardly the most dangerous situation that City have contrived in this long period of pressure since the second half began. Guardiola’s side are looking properly slick.

Minute 54: Jesus and Zinchenko combine to feed David Silva as he’s running into the box through the left channel. It’s a great chance but Silva shoots weakly and McCarthy easily saves.

Minute 56: Cancelo whips it in and Jesus jumps and takes a swipe at it with his left foot. It’s on target but it is deflected for a corner. Southampton win the first header but City win it back; Cancelo finds David Silva in the box, who passes it to Jesus. Jesus displays some nice footwork to get a shot away but it’s through a crowd and James Ward-Prowse hoofs it away.

Minute 57: Remember when Ward-Prowse cleared the ball a moment ago? City retrieve it in their own half and Fernandinho plays in Zinchenko with a long diagonal ball, who finds Bernardo Silva around the D. Silva’s shot is on target, but blocked. It’s not the greatest chance but it’s a decent one - they’re really starting to add up.

Minute 59: KDB is on now. Foden as well. Mahrez and Sterling off.

Minute 65: Bernardo Silva with a blocked shot from a good area after being played in by some interplay between Zinchenko and de Bruyne on the left-hand side. He hits it with some power and it has to go down as a decent effort. I’m of the opinion that City should sell Zinchenko to like Valencia or Schalke or something because he’s incredibly hard to tell apart from de Bruyne on the pitch and I find it confusing and irritating.

Minute 66: Zinchenko puts in a dangerous cross, Jesus does reasonably well to meet it but his header is too central and doesn’t have enough power on it. The resemblance between Zinchenko and de Bruyne is already beginning to piss me off.

Minute 74: De Bruyne takes a snapshot on the turn. It’s off target and not a very good chance.

Minute 77: De Bruyne shoots from just outside the box on the right-hand side. He’s certainly capable of scoring those! Alas, it is blocked by Jack Stephens.

Minute 79: Fernandinho has a long-range shot blocked, which technically counts as a shot. Immediately beforehand, however, City have gotten the ball into the center of the box under the control of David Silva, who very nearly slips in Jesus. Maybe Silva should’ve shot there?

Minute 82: Yet another blocked shot, this time David Silva. Foden crosses but Stephens beats Jesus to it. He is only able, however, to get it away to the feet of Silva. In the end, the shot is pretty tame.

Minute 84: Laporte has a crack at it from way out. He kicks it very hard but it’s very much not on target.

Minute 90+3: De Bruyne plays one over the top for Bernardo Silva to try and volley. It’s not an easy one and he doesn’t quite reach it.

Minute 90+5: Another excellent cross from Cancelo, maybe the man of the match. McCarthy flails at it but can’t quite grab it, it lands at the feet of Bernardo Silva, who only ends up kicking it against Stuart Armstrong for a corner. It could well have been an own goal.

Minute 90+7: City win a free kick right outside the D for a shirt-pull on David Silva. This is in de Bruyne’s range no doubt. He hits it straight into the wall. That’s it for the match. City’s players look a bit nonplussed but mostly apathetic.

As mentioned, City’s failures this season are largely down to defensive issues rather than a truly chronic inability to score good chances (even after this strange match Understat has them only a few behind their expected goals scored). Guardiola’s side have dropped points in a few games that they’ve conceded two or more goals in - against Wolves, United, and champions Liverpool they were undone by genuine defensive frailty, and in the case of their 2-2 draws against Palace, Newcastle, and Spurs they were rewarded reasonably well for their attacking performance, but this was nullified by a spectacular conversion rate of mediocre chances by their opposition. It would be easy enough to look at a result like the one against Southampton in the context of City’s season overall and figure that it must represent some aspect of the kinds of issues that have afflicted City all season, but this is simply not true - the match is not the first one in which City have been frustrated and failed to turn territorial and possessional dominance into a win, but the manner in which it was so is fairly aberrant.

If one were to have the agenda of proving that City are either bad at finishing or unlucky, then they would say that while this was the most dramatic, perplexing, and frankly funny attacking underperformance by City this season, it was not exactly the first. City’s two-goal haul from seven xG home and away against Spurs must rival this game in terms of the capacity to raise Pep Guardiola’s blood pressure, and in losses to Norwich, Wolves, and Man United, City were somewhat under their xG, which might lead to the impression that City have been adversely affected by finishing problems this season. In truth, away to Spurs and away to Southampton are the only matches that look truly improbable or beyond the pale of xG in this regard; under- or overperformance of xG here and there is entirely normal and I’m not going to look at City running about a goal under their xG (as they did against Norwich or against Palace, among others) as evidence of a serious finishing problem anymore than their swing of a goal over xG against Arsenal, Burnley, or Brighton as evidence that they are a side who tend to finish above their xG. When dealing with a side who shoot the ball from good positions as frequently as City do, there are bound to be some slightly odd-looking results against xG. In the case of CIty’s 0-2 loss at home to Wolves (in which they generated 1.9 xG in attack) they could be painted as a touch unfortunate not to score a go-ahead goal, but I think this would be too sympathetic towards Guardiola and his players; having actually watched that match it was clear that City’s attack was too narrow, with Ilkay Gundogan and the Silvas constantly attempting to unpick Wolves’s low-block. Without the presence of Leroy Sane to tear a defense as good as Wolves‘s out of shape, the composition of City’s squad meant that they were reliant on utilizing quick passing combinations and through-balls to nick a goal - their efforts were futile at creating anything more than half-chances, and they deservedly lost that match.

My point with the Wolves example is simply that most of City’s truly mediocre attacking performances this season should not really be grouped with the Southampton match to illustrate a broader pattern of bad finishing or bad luck. Could City’s attack have conceivably generated a couple more goals in relatively close matches against Wolves, Man United, and Liverpool, among others? Undoubtedly, if they’d had their shooting boots on or faced lesser keepers. But these were not evidence of endemic bad finishing or bad luck, simply variance going very slightly against City. In contrast to these, most of City’s ultra-dominant performances in terms of possession and shots have in fact resulted in wins by margins of three or more goals, including against Brighton, Villa, West Ham, Arsenal, Palace, and Watford, the last of whom City managed to score against eight times. By and large when City are at their best going forwards they are a goal machine, and the away-losses against Spurs and Southampton are exceptions to this rather than indicative of any larger pattern. Are City unfortunate to have suffered these two defeats in a single campaign? Probably. But for us to read into them some underlying problem with City’s attack, it would be necessary for them to clearly represent some larger pattern rather than going against the grain of the trend that when City create loads of shots, they tend to score loads of goals. 

One possible explanation for City’s profligacy in the Southampton match would be that the club’s all-time top-scorer Sergio Aguero was out with an injury and his understudy Gabriel Jesus was not up to scratch. Jesus is, in fairness, a player who for all his link-up play and exuberant pressing has tended to finish a few goals below where xG thinks he “should” in a given season. I won’t speculate or offer any analysis as to why he exhibits this tendency beyond that he doesn’t quite show  the kind of attribute that Aguero does to use strength and balance to hold defenders off, turn quickly, and shoot with pace. Jesus is still a reasonably prolific scorer for his club and, it should be noted, has a better strike-rate for his country than Aguero does. Aguero also started at White Hart Lane, which was arguably an even worse finishing performance, so it’s not like when he’s on the pitch City don’t ever miss good chances. The question of finishing skill is a nebulous one, but it would not be totally unfounded to say that while Jesus has the intelligence and agility to get into good goal-scoring positions, his finishing lets him down more than it would most comparable strikers. Is that evident in the shots he missed in this match? Not really, they weren’t easy chances and he should be given credit for getting them at all rather than chastised for missing them. Guardiola has reluctantly relied on Aguero quite a bit during his time at City, presumably because the Argentine‘s gratuitous finishing skill makes him a reliable enough goal threat that even when CIty’s collective attack is not quite at its fluid best he can still get them on the scoresheet. At the same time, City do not rely on Aguero in the same way that, say, Barca rely on Lionel Messi -  Aguero also scores a smaller proportion of City’s goals than Robert Lewandowski does Bayern’s.

The possibility that this was an excellent Southampton defensive performance also comes up. In a sense this was quite Atleti-like from the Saints, befitting their red-and-white-striped shirts; Hasenhuttl’s team didn’t simply park the bus, they were incredibly aggressive in putting pressure on the ball when it arrived in particular areas while retaining a compact 4-4-2 shape. It was reasonably effective at unsettling City as they tried to move their buildup into Southampton’s third, and more importantly a very opportune moment of pressure produced the game’s only goal. And like Atleti, Southampton relied on their back four and goalkeeper to block any half-chances that City could create and win headers - Jan Bednarek and Jack Stephens are not peak Miranda and peak Diego Godin, and Alex McCarthy certainly is not Jan Oblak, but they did their jobs and “put their bodies on the line”. A lesser goalkeeping performance than the one displayed by McCarthy definitely would not have kept City from scoring. At the same time, this was not an entirely successful Atleti-style defensive “masterclass” given that City regularly did take fairly high-quality shots from inside the box. A good comparison might be Atleti’s second-leg performance against Liverpool in the Champions League, or their second-leg match against Bayern in 2016, both of which required a lot of last-ditch defending, goalkeeping, and poor finishing to stop quite rampant attacking performances. It’s entirely possible for a defensive effort to be praiseworthy in terms of its “grit” and so forth even if it results in some decent chances that require saving or being missed. That’s essentially what this was. Wolves and United have genuinely stopped City from scoring by keeping them from creating good chances this season, but that explanation isn’t really available in this instance.

There are other, more speculative explanations for why City failed to score any of their chances this match. Rather than blitzing Southampton from the first minute City knocked the ball around and played their way into the game methodically - they only mustered their first meaningful attempt after going behind! From this we might well infer some psychological factors in play - it could be that City’s attackers were anxious when they were taking their shots because they knew that they had a deficit to make up, and Southampton’s defenders might have been especially committed to challenges and blocks because they had a lead to protect. These kinds of dynamics are a part of the game, sure. Still, CIty’s players are professionals who have won five domestic trophies in two years, and this was not a high-pressure match, it’s hard to imagine that they were so demoralized or anxious that every one of them simultaneously lost their edge in front of goal in these circumstances; likewise Southampton’s defenders and keeper put in good individual performances, which they did look motivated for, but there are a lot of defenders with all the motivation and confidence in the world who aren’t able to block the kinds of shots that City took in this match. There’s probably some kind of a “game state” explanation where Southampton had more bodies in the way of City’s shooters because they had a one-goal lead and reverted to a defensive plan accounting for that, but all of that is unconvincing when one actually goes back and watches the match - if Sterling or Mahrez score with their presentable opportunities midway through the first half it’s a different game, and when one breaks it down to that level of specificity it seems crystal clear that the match largely didn’t hinge on any kinds of factors like that. As they are wont to do City created plenty of moments in which the match could very conceivably hinge in the direction of a comeback win for themselves, but in every instance it hinged further in the direction of a frustrating evening. That’s football.

Bernardo Silva, David Silva, Kevin de Bruyne, Riyad Mahrez, and Raheem Sterling all took shots in this match. All have been among City’s major goal threats over the past few seasons, and none have looked at all like “bad finishers” either to the eye or based on xG. None of them missed sitters of the “harder to miss than to score” variety but the Silvas in particular had straightforward chances that on another day they score. In the aforementioned Wolves game City fed on thin gruel when it came to shots from central areas in the box; here they had quite a few. This was far from the best performance that they have put up under Guardiola, but it was a good one. There is really no compelling explanation for the result beyond goalkeeping and finishing variance. This seems problematic if one views it as part of a much larger pattern because it seems completely, utterly inexplicable that variance of this kind could affect a single team repeatedly over an entire season. In reality, while City have undoubtedly had some wonky results over the course of the campaign, the only one that bears a particularly strong resemblance to this one is the aforementioned Spurs match. It is easy to watch a game like this and view it as summing up City’s season, but this would be something of an illusion. City’s season has been frustrating and disappointing, that’s for certain, but this was frustrating and disappointing in a relatively novel way.

[tactical-substitution]

The Konundrum of Kai Havertz

One of the principle things that differentiates a footballer from a writer, artist, or philosopher, relates to the significance of questioning and answering. Whereas in the latter examples, the very best ones tend to be characterized by the capacity to create puzzles and contradictions that invite further thought and insight, truly elite footballers are different; their presence on the pitch feels like a certain “this is just the way it is”. It can be surprisingly difficult to tactically analyze the Messis, Xavis, Ronaldos, and Lahms because the best footballers in a given generation excel so completely at their assigned tasks on the pitch that there are few questions left to ask. It is the great-but-problematic footballers who ellicit further reflection. It often ends up being the Sneijders and Goetzes and Balotellis who prompt fans to think and debate about exactly what they were good at, and why this was never enough to be “world class” footballers on a prolonged basis. If the former category could currently be said to consist of players like Kylian Mbappe, Virgil Van Dijk, and Kevin de Bruyne, it is unlikely that most football fans won’t already be thinking of someone who meets the latter criteria – a Romelu Lukaku, say, or a Sergej Milinkovic-Savic. It is obvious when watching some players play that there are things that they are capable of that few others would try, and yet actually fitting these players into the squads of clubs among the five or ten best in the world proves difficult. It is this space between manifest skillfulness and tangible skillset that would allow them to play at the absolute highest level that tends to illustrate what is required of the players for whom no such deficit exists, and creates a basis for scouting and player analysis at the level of potential Champions League winners.

Among younger players, little doubt exists as to the sufficiency of Kylian Mbappe or Jadon Sancho should a European giant wish to sign them – in virtually any tactical circumstances their technical and athletic gifts are enough to wreak havoc on any defense. These are the kinds of players who will set clubs back and arm and a leg, and be worth it. In this transfer window, there is perhaps no better example of a “great-but problematic” player attracting serious transfer interest than 20-year-old German international Kai Havertz, currently at Bayer Leverkusen. The young attacker has been seriously linked with the likes of Liverpool and Bayern Munich for months, though Chelsea have ostensibly lept to the front of the queue, having apparently only started on their spending spree with the acquisitions of Hakim Ziyech and Timo Werner. Barcelona, Real Madrid, PSG, Juventus, Man City, and Man United have all been linked at least somewhat credibly with attempts to woo the young German. Whether he moves in this transfer window, and if so where, could come down to any number of factors: Will Bayern prioritize his signature because of his nationality, and will the player feel the same way? Will the desire the work with a celebrity manager like Pep Guardiola or Jurgen Klopp be an incentive to push for a move to a particular club at a crucial point in the young player’s development? Maybe most germanely, who will actually be willing to spend the money should Leverkusen hold out for his buyout clause? In the cases of Sancho or Mbappe, nearly any fee or wage would be rewarded by a signing that is as close to a “sure thing” as exists; with Havertz, the risk seems to be far greater. Knowing why this is the case will be crucial for any signing club if they wish to mitigate that risk, and to not ensure that they have a player widely derided as a “misfit” or “flop” on their hands collecting high wages and attracting negligible transfer interest in two or three years’ time.

What makes Havertz a risky proposition relative to other putatively world-class youngsters is not a matter of talent or lack thereof – go watch a YouTube highlights video if you doubt he’s a marvel – but rather one of style and skillset. Analysts have observed that Havertz’s position could be thought of as a fairly orthodox and old-school number ten, which is a problem given that the clubs interested in him don’t necessarily play with tens. It is not simply that any one club incidentally does not currently play with a playmaker “in the hole”, but rather that the tactical dynamics of modern football have crowded such players out. Just look at the (mis)treatment of Mesut Ozil or Philippe Coutinho by fans compared to the universal adulation given to the likes of Thiago or Marco Verratti for their more “complete” midfield performances. Champions League holders and runaway Premier League leaders Liverpool are generally noted to play with a trio of “workhorse” midfielders supporting their adventurous fullbacks and explosive attackers, and have improved their fortunes dramatically since jettisoning Coutinho to Barcelona, while the player looks like a black sheep wherever he goes in spite of being a much more skilled and “watchable” player than current Liverpool midfielders like Jordan Henderson or Georginio Wijnaldum. Barca themselves were the best club side in living memory on the basis of the genius of midfielders like Xavi and Andres Iniesta on both sides of the ball, and were frankly foolish to regard Coutinho as a “replacement” for all of the things that Iniesta did on the pitch. Top teams attack, defend, and press as an organized unit, and it is hard to see a role for a “free” playmaker who cannot also cover spaces and defend individually in midfield, participate in the buildup of possession, and generally act as a multi-functional cog in a tactical system, or else play as out-and-out forwards stretching play, battling with defenders, and creating and scoring goals. Therein lies the rub for Havertz, who for all of his flashy ability does not really profile as the kind of player who can fill any of the roles in a truly modern football side that looks to dominate with and without the ball, at least without some development on his part, or some shoehorning and accommodating on the part of the team.

What it means to play as a number ten can be variable to different contexts, ranging from deeper-lying playmakers of the ilk of Carlos Valderrama, to creative attackers who would tend to play off of strikers or even as “false nines”, a la Francesco Totti. Havertz is much more in the latter mold, and in fact Totti is a decent comparison to his style of play. What Havertz is truly great at is using his balance, ball control, precise shooting, and passing range to conjure up “moments of magic” – high risk, high reward actions in and around the penalty box that if they work are very likely to lead to a good goal-scoring opportunity. In addition, his height and heading technique make Havertz an aerial threat if the ball is crossed to him. So far, so good – many analysts and pundits regard the scoring of goals as a tactical end worth pursuing. The problem, then, is that Havertz isn’t terribly good at doing much else. His one-on-one defending is nothing to write home about, he lacks the explosive pace to beat all but the slowest and most injured fullbacks if he ends up in a wide position, and he rarely involves himself in buildup play closer to his own goal. It is highly unlikely that he will get much faster, though he may well gain some physical strength and tactical intelligence – nevertheless, simply expecting the player to become a more solid defender or exert a more metronomic influence as he ages is frankly a gamble. If a big club decides to go in for Havertz they should be able to fit him into their plans commensurate to the amount of money they invest in him, and they should be able to do so now rather than in some ill-defined future where his game has become more balanced and less deficient. Accordingly, the thorny task around Havertz is determining whether his game, more or less as it currently exists, could fit into those of any of the superclubs he is linked with.

The most conventional possibilities for Havertz’s future are worth considering, and roughly align with the roles he has played at Leverkusen. He could play as a pseudo-right-winger, though this would necessitate an overlapping fullback or wingback to give width in attack. He is also possibly capable of playing as a withdrawn forward in the mold of Roberto Firmino, chipping in with a non-embarrassing goal tally but also pulling the strings creatively while the goalscoring burden is carried by wingers, a strike partner, or advanced midfielders. Neither of these are totally inconceivable, but unless Havertz irons out kinks in his game and broadens his skillset, they would necessarily pull teammates out of position to fill the gaps he leaves, potentially creating problems elsewhere on the pitch. Top managers quickly notice these kinds of things and are unlikely to be so impressed by his neat touches and controls that they don’t yank him from the starting XI in their quest for tactical impenetrability and balance. Another possibility is that Havertz will play as a “second striker” off of a more traditional number nine for the rest of his career. One system that could facilitate this would be some species of 4-2-4, with midfield areas occupied by a strong double pivot; another would be to position the wingers more conservatively so that the team lines up in more of 4-4-1-1. The former would be pretty absurdly attack-minded, perhaps resembling Pep Guardiola’s 2015-16 Bayern side when Kingsley Coman and Douglas Costa played high and wide, the fullbacks tended to assist the midfielders, and Thomas Muller (a German number ten a decade Havertz’s senior) played off of Robert Lewandowski. When fully functional, that side was mind-melting to watch, and Havertz might strive to emulate Muller’s successful interpretation of the attacking midfield role, with defenders never sure if he had dropping deep to create or darting to meet an aerial cross on his agenda. If Havertz were to play in the latter system, his role might be comparable to that of Antoine Griezmann at Diego Simeone’s Atletico Madrid, or perhaps a frame of reference would be the function of Juan Mata, Eden Hazard, and Oscar for Chelsea under managers like Jose Mourinho and Antonio Conte. The allure of this kind of role would of course be that Havertz is less likely to come across as a defensive liability in a system where virtually everyone is responsible for a great deal of defensive grunt work in one way or another and he is the primary creative outlet, constantly looking to locate himself in a point of weakness in the opposition’s structure, receive the ball, and then make the “magic” happen. Where problems come up is when one remembers how much running around the likes of Griezmann or Oscar always did. Even in a counter-attacking side, even if a player does much more than their fair share of creative passing and danger-creation, they will still inevitably be expected to harass opposition defenders and midfielders out of possession, and Havertz has not yet clearly demonstrated the stamina or the tactical intelligence to be apt to such a role.

One last possibility for Havertz would be to play as a kind of “false midfielder”. He could stand in midfield areas to make up the numbers and aid in a tactical plan to create numerical overloads in central areas, and then when able to get closer to goal, really start to add value. The problem here is that Havertz has not yet shown that he can become a safer, higher-volume passer at even the level of, say, David Silva or Isco. Those midfielders are nowhere near the central midfielders that they are attacking ones, and cannot resolve situations in congested areas through passing combinations at the level of their compatriots Iniesta or Xavi, but they can at least participate in such exchanges without constantly playing catastrophic sideways passes which result in counter-attacking opportunities for the opponent, and thereby justify their presence on the pitch for when they are able to be more useful in the final third. Is Havertz trustworthy enough to do so? If so, he has not yet shown it. With all of these hypotheticals, it is not outside the realm of possibility that the player will simply develop his game through practice, coaching, and the last embers of puberty, and be able to play them satisfactorily, but this seems like too speculative an expectation to ground such a large financial outlay in. Havertz is who he is right now, and should be scouted for his demonstrated skillset rather than based on wishful thinking. If a club is to justify the purchase of Havertz, then they may have to fundamentally rethink the kind of player he is, and re-rationalize how they will fit him into their plans.

What clubs should expect to get should they bid for Havertz, is a kind of very expensive water-carrier. When most football fans hear that phrase they likely do not think of a player of Havertz’s qualities. French attacker Eric Cantona infamously used the epithet to describe his international teammate Didier Deschamps, as a kind of tongue-in-cheek way of belittling his compatriot’s contribution to the collective effort. Other players who tend to come to mind as “water carriers” might be Marcel Desailly or Claude Makelele. What all of these players have in common, besides their nationality, is a particular skillset, which was as narrow as it was well-executed. These midfielders had the match-intelligence to step up and engage opponents at the right times without getting caught out of positions, assisted defenders at risk of being over-burdened, covered for teammates who may have been out of position, were precise in challenges, could effectively mark dangerous opposition players as effectively as a center-back, and were mobile enough to cover ground and arrive at the right time and place to slow down or stop opposition attacks or win the ball back. In possession they were all fairly conservative, prioritizing consolidating possession and passing to teammates in space over attempting any risky passes or skills on the ball. They are greats of the game because they “carried the water” for more flashy teammates who were given more license to improvise and take risks closer to the opposition goal, by performing these kinds of simple actions with great competence and consistency. Clearly this does not sound like the kind of player that Havertz is, and in fact quite the opposite, but what if there is more than one way to carry the water?

If the main impression that most fans have of Cantona’s famous phrase is that it describes the duties of a defensive or holding midfielder, there is perhaps another interpretation of it that should be considered, which has more fidelity to his intentions with the quip: a “water carrier” is a player who performs a narrow range of mechanized actions to support the team, in contrast with teammates whose roles entailed doing more different things, in a greater variety of situations, in order to really unlock games and break down an opposition. In Cantona’s time this was arguably a fairly accurate rendering of the dynamics of how a footballing side functioned – midfielders and defenders tended to keep a compact shape closer to their own goal, and tens, wingers, and center-forwards were given the task of breaking down those compact structures with “moments of magic”. In such a (simplified, admittedly) context Havertz would be an absolute star with his skillset, as he would constantly have the game in front of him, virtually daring him to find a distressed point in an opposition’s defensive structure in which he could dribble, run into, or find a teammate with a through-ball. There is no doubt that defenders and midfielders would find it difficult to deal with Havertz’s clever runs, deft controls, and overall skillfulness for approximately half of a ninety-minute football match, and the onus would be on his “water carrier” teammates rather than him to do much of the serious running and remain positionally alert. Perhaps in such a situation Kai Havertz would truly be living high on the hog. Unfortunately for the young German, he was not even born at the time that that footballing milieu existed, and so must contend with the landscape of 2020.

When one watches the football sides of Guardiola, Klopp, Tuchel, Pochettino, or Sarri, it is pretty abundantly clear that the players are not divided into conservatively-positioned grunts and attacking artists running around at the will of their self-governing genius improvising ways to foil the defensive goons. It is a cliché to describe top-level sides in the modern game as “fluid” or praise their “discipline” and “organization”, but there is a good reason for that. Modern approaches are based on using all ten outfield players, and increasingly the goalkeeper, to circulate and advance the ball into positions of overloaded strength in possession, and stymy and harass opponents out of doing any of that out of possession. The number nine is the first defender, and the goalkeeper the first attacker, and all of that. It is somewhat difficult to see a place for a solitary magician like Kai Havertz in all of that, as modern footballing sides tend to rely on the industry and intelligence of their forwards to keep opposition defenders from building up attacks, and don’t tend to rely quite as heavily on one or two especially flashy players to break down defenses as sides of the past. For a modern superclub, an attack might look something like a left-sided midfielder and left-winger forcing an opposition defender to play an errant long pass by swarming them and giving them little option, a left-back retrieving the ball and playing it to the holding midfielder, who consolidates possession with a quick interchange with the right-back, allowing the right-back to engineer a free path close to the opposition penalty area where they can cross to whichever attackers or midfielders are able to plausibly compete for an aerial ball with the opposition defenders. In scenarios of this kind, Kai Havertz is unlikely to play much of a part, and if he does is unlikely to be doing anything that a less costly player with a different skillset would be unable to. This is before one considers his (relative lack of a) defensive contribution. In these kinds of tactically sophisticated, high-tempo “gegenpressing” encounters, Havertz begins to look like something of a flat-track bully and an anachronism rather than a superstar.

How, then, is Havertz to fit in to a modern footballing side, if at all? In the above scenarios, and indeed in the squads of big clubs like PSG, Juventus, and Bayern, midfielders (and increasingly fullbacks) are expected to have broad skillsets, and to be able to move and progress the ball in a variety of ways and to defend actively and passively, and indeed academies across the world are producing players who can “do it all” without any “master of none” caveats. Take potential Havertz destination Bayern for example: Austrian defender David Alaba, who has spent the bulk of his club career at left-back but was groomed for a kind of hybrid center-back/midfield role by Guardiola, and plays as an attacking midfielder for his country, plays as a center-back playing incisive long and short passes to start attacks, and uses his anticipation and speed to expertly mark opponents. Former right-back Joshua Kimmich has played in central midfield this season, and has arguably been the best player in his position in the entire world. Young Canadian winger Alphonso Davies has deputized at left-back, and, like Alaba and Kimmich, has been a strong candidate for the most effective and complete player in the world or at least the league in his position, despite it not actually being his position, using his lightning pace to monitor an entire side of the pitch, overlapping intelligently in attack and showing excellent reading of the game to make vital defensive clearances and tackles, all while looking frigidly cool in building up possession in conjunction with his more experienced comrades. These are as much the stars of Bayern’s juggernaut team as their vaunted forward players, along with the defensive likes of Niklas Sule, Benjamin Pavard, and Jerome Boateng. Whichever system Bayern choose to play, it is unlikely that Havertz would be their most important player or even particularly close despite playing in a role that would render him the traditional “star player”. What would the young German mark himself out as, then, were the Bavarian club to take a punt on him?

In the tactical schemes employed by top clubs of the 1990s, elite attacking midfielders of the ilk of Roberto Baggio, Zinedine Zidane, and Michael Laudrup were not only the “stars” of their club and national sides in terms of press attention and shirt sales, but were also typically the players who would quantitatively have a “starring role” in the sense of touching the ball with greater frequency and significance than teammates. Over the last two decades and under the influence of the “positional” approaches of Marcello Bielsa, Louis Van Gaal, and Guardiola, and the “counter-pressing” philosophies employed by Klopp, Roger Schmidt, and Ralf Rangnick, deeper-lying, less directly creative midfielders have emerged as the “protagonists” in matches where the majority of clubs prioritize ball retention in safe areas, and use structured possessional routines to both keep a compact defensive shape and manipulate the ball into dangerous areas. On and off the ball it has been midfielders such as Sergio Busquets and Mousa Dembele and, increasingly, defenders like Trent Alexander-Arnold and David Alaba who have not only spent the most time directly controlling the play of games, but also had to use the widest array of dribbles, passes, tricks, and controls to keep and manoeuvre the ball, and this has been by design. Even many smaller clubs have come to adopt this style of play, but, most relevant to the immediate future of Havertz, every title-chasing club across Europe with whom he has been linked play this way. Gone are the one-dimensional “water carrier” defenders and holding midfielders, and in their place a generation of deeper-positioned players with the skillsets of traditional number tens (many of whom in fact played in that role at youth level or earlier in their playing careers) have emerged. At the tip of the spear pace has been the attribute that most preoccupies top managers, whose obsession has been with using the coordinated movement of attackers to provide passing options and open up spaces for one another and for advancing midfielders and wingbacks or fullbacks. These attackers have found themselves in the role of “water carriers” for their more cultured teammates further back on the pitch, performing simple actions in a relatively mechanized way, but doing so at such a high level that the new breed of holding playmakers are able to use their broader skillsets to move the ball into areas which have been given situational tactical significance and danger by their water-carrying forward comrades.

If the modern “water carriers” are the quick and tricky attackers whose speedy movements create the chaotic conditions which their teammates exploit, then how is a player with little natural pace or explosive acceleration to distinguish themselves? This is the scouting question that lingers over Kai Havertz – his actual skillset looks quite narrow against a frame of reference where playmakers are expected to do their work in more congested and treacherous midfield areas, but if a club side were able to exploit the young German’s demonstrated capacity to pass creatively in and around the penalty area to its fullest extent, it might well be enough in and of itself to justify his presence on the pitch. The problem is that it is not immediately obvious how his skillset and limitations could be accommodated by the tactical schemes employed by the likes of Guardiola, Klopp, and so forth. In the case of a true two-way midfielder or an explosive winger, it requires little imagination to see how they would be “plugged in” to the kinds of tactical systems used by Man City or Liverpool – the dynamics of these systems are calibrated to balance the extensive skillsets of the star players against the intensive skillsets of the water carriers, albeit in a markedly different ways from the suberclubs of the 90s. In the case of Havertz, it is clear that he would end up being a “water carrier” in a limited role were he to wind up at either club, but decidedly murkier whose water he would be carrying, and what kind of intricate tactical scheme of delegating tasks would mediate this balance. Managers, scouts, and pundits understand the scouting question for an old-school defensive midfielder – will this player’s ability to mark opponents and plug gaps carry the water left at that club by its existing, defensively cavalier squad without creating an awkward stylistic disjunction? The same kind of scouting question will have to inform clubs as they contemplate allocating a large chunk of their budget towards Havertz – in what way can the existing or prospective squad play on and off the ball such that Havertz’s superior skillset around the penalty area will free up teammates to do all of the other important work which they are more suited to without having to worry about scoring and assisting goals as much as they might otherwise have to?

Such a question is radically different from the orthodox perspective from which clubs (as well as football journalists, fans, etc.) tend to approach scouting a “star” player like Havertz, but its pertinence is a consequence of the style of modern football. As Casemiro is a guaranteed starter at Real Madrid because his prodigious volume of tackles and interceptions mitigates his pedestrian on-ball skillset and “frees up” his more expansively-passing teammates, big clubs eyeing up Havertz must consider whether the routineness with which his movement, vision, and skillfulness make the difference in the final third can be utilized as a similar kind of mechanized action to Casemiro’s defensive interventions; such a question is not simply a matter of whether or not the scales can be balanced by such and such a teammate who performs an equal number of opposite actions, but a rather more dynamic one of what types of things a team’s midfielders, defenders, and strikers might suddenly be able to do on the pitch if they are safe in the knowledge that Havertz is reliably carrying the water around the penalty area. This is the kind of complex tactical question that elite managers and sporting directors get paid eye-watering wages to attempt to solve, though it may well be intractable to the point where Havertz is virtually fated to wind up at a club like Valencia or Leicester City in five years’ time. It is also a question with a crass quantitative dimension – if Havertz is only creating and scoring a few goals per league season then the truly big clubs needn’t ponder his unbalanced skillset at all, whereas if he’s directly responsible for two goals every game then all other more nuanced questions become moot in a much different way. In reality, Havertz is in double figures for combined expected goals scored and assisted in both of the last two Bundesliga seasons, and has gone one something of a finishing tear that has made him look like a seriously prolific goalscorer, but he is nowhere near the “get him at any price, figure out what to do with him later” levels of Messi or Neymar, or even Eden Hazard in an average season. Havertz is a seriously tidy-looking attacking midfielder whose flashy skills have led to tangible goal contributions which are nothing to shake a stick at, but he has not as of yet demonstrated that he can shoulder an attacking burden commensurate to the defensive one carried by a Casemiro or an N’Golo Kante.

Scouts, managers, and analysts will have to squint hard at the data and footage, and figure out a way in which Havertz’s skillset can be made to carry a little more water than it currently is - his height and precise heading technique, for example, could probably be exploited more than they are currently being. But this kind of conundrum casts a fog of doubt on the notion of splashing a hundred million Euros on the player – sure, he’s nice, but shouldn’t that kind of a cash outlay mean we don’t have to think so hard to figure out what to do with him? You scour a wine store for half an hour looking for an obscure Greek red on a clearance sale and pair it with precisely the right meal to get the most value out of your wine-buying dollar, but if you go and squander your money on a seventy-dollar Barolo you kind of expect that you’ll open it up and it will just be good – otherwise what’s the point? None of this is to say that Kai Havertz clearly isn’t worth whatever fee and wages a club ends up parting with to secure his services – he can do all kinds of exquisite things with the ball at his feet (and head) that look like the kinds of things that players do in the highlights footage from Champions League and World Cup knockout games. He can do, and has routinely done, the kinds of things that decide those kinds of games, sometimes against the kinds of defenders who try and stop a player from doing them. But rather than looking at him as the next Baggio, Totti, or even Muller, big clubs across Europe should consider the sense in which Cantona belittled Deschamps for doing such a narrow range of mundane things, and stick to the mantra that if they end up buying a water carrier, it is the rest of the squad and the manager who will end up carrying him. If Havertz can end up doing as few things as well as Deschamps did, he will turn into a footballing legend like the current France manager, but if he doesn’t turn out to live up to the “as well as” part then he will only ever be the kind of player who fans and pundits describe as “fun to watch, but limited.”

If the phrase “it is the water carrier who is really the one being carried” doesn’t evoke the image of a white-haired Shaolin monk in a Shaw Brothers film or the sound of a bong gurgling in the bedroom of some philosophy undergrad, then maybe a more traditional brain-teasing dialectic will do the trick – “the more that things change, the more that they stay the same.” When one frames Havertz’s skillset as being that of a traditional number ten, he is cast as an anachronism. When football fans of a certain age hear the phrase “number ten” they are likely to think of big 90s Serie A clubs, and the likes of Baggio, Totti, Veron, Zidane, or Riquelme. Footballing reactionaries sick of the ultra-disciplined grimness through which a club like Liverpool can optimize small advantages and secure a string of one-goal victories on a march to domestic and continental glory might well fantasize about a career path wherein Havertz loafs around at the big Milan and Rome clubs and Fiorentina, playing in a way that gets described as “languid”. In this scenario none of these clubs ever hire a “modern” coach or sporting director, or at least nobody more progressive than Sacchi, and the footballing landscape in Italy is fixed to a particularly simplified representation of the league two or three decades ago. He might even be able to show up drunk, as many football writers seem to insinuate that Andrea Pirlo did when they make gratuitous references to his wine connoisseurship as some kind of synecdoche of his “elegant” and “old-world” style of play. All joking aside, there is a credible case to be made that Havertz’s skillset make him something like a genuine anachronism – he’s likely fitter and certainly has more video analysis sessions under his belt than the old school number tens, but this hasn’t ironed a marked stylistic resemblance out of him. Where this begins to look like some kind of big conundrum is when one looks at where he plays, and who he plays for: Havertz has a starring role at an exemplary progressive, data-driven, modern pressing football club, stewarded by a well-regarded “gegenpressing” manager with stylistic roots in the Holland/Ajax “total football” philosophy.

If Havertz were to play as a “traditional” number ten, more or less singlehandedly acting as a creative outlet in an otherwise defensively rigid unit then he would likely end up under a manager like Mourinho or Simeone, playing the central playmaking role in a 4-4-1-1 or 4-2-3-1 system. It is possible that he would wind up in a system with a three-man defensive line and wingbacks behind him were he to play under the likes of Nuno or Antonio Conte, just as the number ten playmakers in football’s bygone era may well have played in both back-three and back-four systems, but like those players he would expect to play a creative role in front of a well-drilled unit that no one would describe as “fluid”. At Bayer Leverkusen, Havertz has played a non-negligible number of minutes as the team’s sole attacking midfield, in a “free eight” role (similar to the “false midfielder” roles of Kevin de Bruyne and the Silvas at Man City) alongside Julian Brandt, as a central striker or false nine, and in a pseudo-right wing role, and has generally looked very good wherever he has played while never having a hugely different role or set of jobs on the pitch or displaying a different skillset more suitable to the position which he is ostensibly playing. He looks like a classy player wherever he plays, but he always looks like Havertz. This positional carousel is operated by manager Peter Bosz, who has set the team up with back-three, back-four, and back-five defensive systems and rotated his midfielders and attackers significantly, giving minutes to all members of Leverkusen’s impressively deep squad. Bosz seems determined to extract serious contributions from his entire diverse cast of players, and has demonstrated a great talent for devising tactical plans that allow basically any combination of them within reason to be on the pitch at the same time and function according to his broader tactical principles. Havertz will be moved around to accommodate this or that attacking partner, but rather than shining by playing the position in an orthodox manner, much credit should go to the manager for setting up the team in such a way that they are collectively able to engineer the kinds of situations in which Havertz is able to do Havertz stuff.

On paper, the fluid and modern tactical schemes employed by Bosz at Leverkusen could hardly resemble the rigid systems that brought success to managers like Lippi, Trapattoni, or Hitzfeld less, but in a very real sense he has simply found much different means to the same end of accommodating his star playmaker. For the number tens at the superclubs of the 90s, positional freedom and the license to move to where they could do the most damage with the ball was a key condition that allowed them the autonomy to control attacks. At Leverkusen, Havertz similarly tends to “go where the action is”, insinuating himself at the center of counter-attacks and picking up pockets of space in which to pick apart deep-lying defensive blocks when his team have established stable possession in the hopes that a teammate will find him so that he can execute the kinds of high-risk manoeuvres which he is so good at. The boy is not going to drag a full-back very far with an off-the-ball run or physically intimidate an opponent into coughing up the ball in a dangerous area, but let him roam around looking for ways to solve attacking problems with the ball at his feet and he’ll eventually figure something out. In previous footballing contexts, players of Havertz’s ilk would accomplish this with the help of a tight-knit lineup behind them playing in, let’s say, two banks of four, with one or two forwards making attacking runs for them to pick out. At Leverkusen, as in hyper-modern Bundesliga-standard football generally, Havertz does not have as fixed a formation supporting him, but his teammates are essentially trying to accomplish the same kind of support structure as were those of Zidane or Veron, but simply doing so in a more sophisticated and convoluted way.

It makes a certain kind of intuitive sense that in a footballing milieu where the significance of pressing and collective actions on and off the ball have been elevated, and successful sides are more thoroughly drilled than ever with the assistance of video analysis and even the literal use of drone footage, the high-level task of having a bunch of unit of outfield players supporting one “star” player would require a more positionally fluid set-up than the traditional formations. In fact, going back to the earlier discussion, it is not difficult to see how the “stars” would become marginalized and the players tasked with the increasingly byzantine task of freeing them up would increasingly take up the mantle of being their teams’ most significant contributors in this context. At any rate, when one watches Leverkusen play, it becomes clear that apart from Havertz’s colleagues in the attacking line playing in such a way as to maximize his options, the contributions of the players behind him on the pitch are absolutely crucial to his capacity to play the way that he does. Leverkusen have traditional defenders and hard-tackling, orthodox midfielders, but no mere “water carriers”. Bosz often plays with three players in the center-back position, but is disinclined to play three out-and-out center-backs, often preferring instead of play full-back Wendell or one of the identical-twin midfielders Lars and Sven Bender in the defensive band – like Bayern manager Hansi Flick, he clearly considers recovery pace and passing ability to be as important as the ability to win headers and make goal-line clearances in his defenders and is willing to play squad members “out of position” to accomplish this. Full-backs or wingers are often played in wide positions as is customary, but Bosz has shown a habit of playing four true central midfielders in his midfield line, giving a clear window into his tactical outlook and into his strategy for getting the best out of Havertz. Julian Baumgartlinger and Charles Aranguiz are the two most defensively-minded of Leverkusen’s midfielders, with Nadiem Amiri and Kariem Demirbay more obviously skillful technicians when they are tasked with making a creative pass or keeping the ball under pressure. All of the club’s primary midfield options, however, are well-rounded in their skillsets, in the same way that the defensive line are expected to be. All of Leverkusen’s midfielders are instructed to press in unison, play passing combinations, drop into the defensive line, and join the attack when the situation calls for it. When seven or eight players can collectively work at a high athletic and technical level to provide structure, make defensive interventions, create adequate spaces in wide and central areas, and so on, it becomes easy to see how a player like Kai Havertz is able to thrive.

Arguably the most analogous current side to Bayer Leverkusen playing at a high level are Atalanta Bergamo, managed by Italian veteran Gian Piero Gasperini. As with Leverkusen, Atalanta focus on attacking far more than they do on defending, and their roaming playmaker-attackers Josip Illicic and Papu Gomez provide reasonable points of stylistic comparison for Havertz. Within Atalanta’s tactical setup, the importance of midfielders Marten de Roen and Remo Freuler being able to undertake a wide variety of unglamorous tasks related to pressing, defending, ball retention and progression, and keeping a collective structural balance through correct positioning cannot be overstated; neither can the significance of the wide-positioned wingbacks, acting as defensive and especially offensive outlets capable of doing the tireless running that the team’s more technically proficient attacking midfielders are unwilling or unable to do. The team must constantly be able to make up the numbers in various areas of the pitch in order to function in passing, defending, attacking the opposition penalty area, and so forth. Illicic and Gomez are reasonably intelligent players, but Illicic in particular is not exactly renowned for his leopard-fast bursts of pace or tremendous workrate. Like Leverkusen, Atalanta play with anachronistic number ten players by screwing with the formula further back on the pitch, and may provide a window into what ought to be done with Havertz in the future. The problem lies in the fact that Atalanta are a “fun” side, but they are far from a “complete” side. The immense ability of Illicic and Gomez has made Atalanta into something of an offensive juggernaut to the point that they are reliable top-four finishers in Serie A despite having far from the fourth-largest budget in the league, but their presence creates a kind of chain of deferred problem that ultimately results in the club’s center-backs frequently chasing back in comical fashion or playing calamitous forward passes because they are isolated from their teammates. Will a team like Man City or Chelsea, whose problems this season have been much more with their defending than going forward, look at Leverkusen and Atalanta and their status as “entertainers” whose matches produce goals at both ends, and decide that Havertz is a prudent signing? It does not seem entirely likely.

Perhaps if a big club is to roll the dice on Kai Havertz, they should look to play with a support structure similar to that employed at Leverkusen, but simply have better midfielders and defenders than Leverkusen. This is probably not the kind of suggestion for which a sporting analyst in a big club’s employ can hope to receive a big promotion, but it may well work. Leverkusen have pretty talented players all over the pitch, but they do not have any one player as truly excellent as PSG’s Marco Verratti, Real Madrid’s Dani Carvajal, or Chelsea’s Jorginho. It is well within the realm of possibility that a club that can afford Havertz can also afford the kinds of players who are so good that they can sufficiently control the aspects of the game that Havertz doesn’t, and allow him to do what he is exceptional at. The issue, given everything previously discussed, is with the proposition of outlaying such a large sum on the player. Why not simply cut out the middleman, sign some midfielders and fullbacks who can collectively do a reasonable amount of goal-creation and goal-scoring without Havertz’s presence on the pitch and be done with it – is there really a high-level system that is truly balanced in which there is really no substitute for a player of his skillset? After all, an industrious and athletic midfielder can create the possibility of a goal with a well-timed run, and a full-back can cross the ball with a meaningful chance of creating a good goal-scoring opportunity.

The scouting question returns to the one of whether Havertz’s gratuitously skillful on-ball ability makes him viable as a “water carrier” for world-class teammates who can do more than just create and score a decent volume of chances. If not, then surely it is the players whose skillsets are more well-rounded than Havertz’s who should be attracting the ridiculous fees rather than the young German. A more tantalizing question might be whether it is feasible for Havertz to do everything he has shown himself capable of doing around the penalty area at Leverkusen, do nothing more, and still have the rest of the team do plenty of solid attacking work in addition to that done by him rather than burdening him almost entirely with making the attack hum. All of this is of course contingent on the rest of the squad being capable of, firstly, feeding the ball to Havertz in the areas where he needs it, and secondly, doing the defensive and ball-progression work that are not really his forte. Any team with serious structural issues in their defensive and midfield areas should be looking nowhere near the youngster as his expensive purchase is highly unlikely to indirectly solve any issues in defense or build-up, and could quite possibly exacerbate them.

All of these questions could reasonably be rendered moot if Havertz simply broadens his skillset somewhat. It shouldn’t be presumed to be inevitable, but it’s far from unheard of for young attacking midfielders to pick up the aspects of the game which are more subtle than getting a shot away or looking for the killer pass. Genuinely top-tier central midfielders like Luka Modric and Andrea Pirlo initially profiled as classical number tens, and it is not too much of a leap of imagination to imagine the German’s balance and passing range being transposed to deeper areas on the pitch. A version of Kai Havertz who shows enough composure to regularly drop back and assist midfielders positioned deeper than him while also demonstrating the same skillset that he already has is basically Kevin de Bruyne. Then again, a version of Michel Platini whose body hasn’t gone downhill athletically over the past three and a half decades and whose reputation isn’t tarnished by a series of corruption scandals is basically peak Michel Platini. A disgusting, greasy broken Hollandaise that was instead emulsified properly by someone who knows how to cook is a delicious sauce. If clubs get too tempted by the prospect of the kind of player that Havertz could be if such and such conditions are met and base their decision to buy him on that, then they are buying the conditions along with the player, and the potential for an expensive mistake is very much baked into the structure of such a valuation.

Have the scouts for Europe’s super-elite considered all of this and analyzed it in greater depth and with more technical resources than has been done in this piece? Without question. Will this ensure that their decision to bid or not bid for Kai Havertz is the right one? Not necessarily. But regardless of what happens with the young player, consideration of who he is as a player right now, who he reasonably could be, and what kinds of conditions must be met for him to end up a success has facilitated a level of reflection on the nature of modern football that would not be the case with other higher-end players. What kinds of scouting questions need a club ask about Raphael Varane, say, if he’s unsettled in Madrid? He’s ridiculously quick, his positioning and decision-making are as good as it gets, he wins his aerial duels. There is no question, go all in for him. What about another 20-year-old Bundesliga sensation, Jadon Sancho? His quick feet, strength, and quick change of direction make him one of the most effective dribblers on the planet, and he’s demonstrated plenty of end product over the past two seasons in Dortmund. If you could use an attacking player and have the money, you go for him, little analysis is necessary. But human agency tends to orient itself towards problems, contradictions, and puzzles, and when a player is as manifestly brilliant as Kai Havertz is without necessarily fitting into any of the truly elite European superclubs, the urge to philosophize his situation takes over. It is this urge which will guide how clubs scout Havertz, and every other player under the sun, if they are to grab themselves a future superstar or at least a useful contributor and not another expensive disappointment.

[tactical-substitution]

Have Liverpool Been Lucky?

At the end of the day Liverpool fans around the world are not likely to be too preoccupied with how the club got into the position they’re in - the Reds are a few points from securing their first league title of the Premier League era and that’s sufficient reason to feel about as gratified as a human being could about football without personally being involved in an achievement. However, there are already grumblings about the role that luck has played in the historic season. At the point when the Premier League was suspended, Liverpool had picked up 82 points from 29 matches, meaning a rate of 2.83 points per fixture - this astonishing total eclipsed even Manchester City’s record from two seasons ago and, coupled with City’s stuttering form and the absence of any credible challenger (or at least a challenger credible enough to keep up with a 100+-point pace), meant that while other big European leagues still had competitive title races of some description at the time that the global pandemic put a stop to spectator sports, any such title race in England’s top flight was purely notional. Quite apart from this, the sheer rate at which Liverpool have won matches and picked up points has given Jurgen Klopp’s side an aura of being a truly historically good side, maybe the best that the Premier League has seen and comparable in quality to Sacchi’s Milan or Guardiola’s Barcelona.

Inevitably, this level of hype has led to a backlash of sorts, with plenty of spiteful opposition supporters cherry-picking certain video-assisted decisions in Liverpool’s favour and calling the team “LiVARpool”; moreover, most “neutral” viewers and even the club’s supporters have noticed the sheer number of times that they have won games after the 80th minute, sometimes with goals that could be described as fortuitous. All of this is to be expected, and well illustrates the appetite of football-viewers for postmortems of even the most settled matches and competitions. Of course, none but the most conspiracy-minded Liverpool-dislikers would suggest that the club should not win the Premier League trophy due to any good fortune that they may or may not have had, and it would be absurd to begrudge them their trophy success given how frequently the club have managed in one way or another to outscore their opposition. Apart from some kind of serious match-fixing scandal there are no legitimate grounds to deny them the title that they’ve earned - football is, as they say, a results business. Why, then, does Liverpool’s supposed incredible luck get brought up so frequently? To what end do skeptics posit that the Reds have been “lucky” if not to deny them tangible success? Clearly, there is something harder to grasp than winning or losing this or that match or trophy that is being disputed by postmortems of Liverpool’s season which would portray them as “overrated”.

In truth, for all that results and trophies are important symbols in football, they are the tip of the iceberg for what makes the sport compelling. It is pointless to watch a match or competition purely to see what the outcome is, and nearly anyone who earnestly supports a particular side will tend to describe the experience of watching a consequential match in which that side is playing as less pleasurable and more distressing than watching one in which they don’t really care what the outcome is. The satisfaction in watching a football match comes from watching two sets of players attempt to execute a game-plan that will get them the result they need, and the myriad dynamics that ensue in the friction between the two approaches. This satisfaction extends to analyzing in one’s own head why such and such a result occurred, and can be further extended into discussions and arguments about the internal mechanics underlying the result - the talents and qualities of individual players, the interactions between opposing tactical schemes, and so on. No one has a theoretical discussion or a heated drunken argument about whether a match ended 2-1 or 1-1, but every football fan will occasionally (or not-so occasionally) attempt to prolong their engagement with a match or resolve the contradictions that it seemed to posit by arguing that one particular side ought to have won a game that finished 1-1. As matches are joined together into historical chains of seasons, campaigns, tournaments, and less formally determinate periods, it becomes de jour to opine about the legacies of particular sides not in terms of whether they deserved to win any particular match, or even in terms of whether they ought to have won trophies, but purely in terms of how good they were.

The disadvantage to the total priority of goals scored in determining the outcome of a football match is that sometimes there is a commonly-shared sense that “the wrong team won” as determined by which side played “the better football”. The advantage, of course, is that matches actually have results rather than being theoretical exercises or treatises whose significance requires any interpretation. In fairness, sports such as boxing and figure skating do rely heavily on subjective, discretionary judgment to determine outcomes, and they get along just fine, but most football fans agree that such criteria would be difficult and vague in football. “Pure goodness” in football is difficult to measure, even subjectively, as there are so many ways for a footballing side to skin a cat and so many factors determining the dynamics and style of a particular side or match.There may be many matches so lopsided in every way except for the final result in goals scored that it would seem intuitively fair to revise the result to reflect the broad superiority of one side, but these are greatly outnumbered by matches where the “obviously” better side wins, which are themselves outnumbered by matches in which each side plays decently enough and both creates and prevents enough scoring chances to potentially win or at least draw. In the latter case in particular few would deny the value of counting each side’s goals in settling matters and progressing things as a competition.

In spite of all this, and especially when interpreting teams across longer periods of time, fans tend to talk about the higher-level qualities of those sides, which cannot be represented by scorelines or points totals. The inability of these qualities to be adequately represented by tangible markers of on-pitch success is clear from the fact that they are often posited in a way that actually contradicts the real-life record as determined by match results by goals scored. It is manifestly obvious that when people watch and think about football they are convinced that it is possible to assess how purely good a team is because they keep trying to do it, but it is equally obvious that doing so is a difficult and perplexing matter from how consistently bad they are at it. Whereas the representation of actual competitive success is so linearly transitive that anyone who doesn’t have severe dyscalculia should be able to work out which team is the most successful given the pertinent goals-scored information, how good or bad a side “really” is at football seems to be necessarily be represented by an interminable sequence of arguments and attempted proofs, many of which are poisoned by the biases and presumptions of those making them.

The most prominent attempt to achieve a sustained and quasi-objective representation of the actual quality of footballing sides has arguably been the application of data analysis models to the ever-larger amounts of information aggregated about football matches. If all of this data is a fairly accurate representation of the quantitative minutiae of matches, players, and teams on its own, attempts to formalize the collection and interpretation of it constitute a mode of argumentation whereby the data is putatively formed and disciplined into representing their qualitative dynamics. It should be noted that the creators of well-known data analytics models in football such as xG (expected goals) have always cautioned against the use of these models to render an impression of what teams may have “deserved” or “earned”, preferring to see them as predictive of how a team will likely go on to play. For all that a predictive data model can’t render about a football match, there are plenty out there which are able to render with a high degree of methodological granularity how much the opportunities a particular side have had to score all-important goals resemble those which have been most historically likely to lead to goals; following from this, and adding up all of these opportunities, an xG model is an instrument for consistently analyzing them within their proper context and subsequently spitting out an impression of the demonstrated capacity for any given side to create and prevent goals.

There are several competing xG models, all continually being refined and improved in terms of their ability to accurately parse, recognize, and render the numerous overlapping senses in which situations are or are not likely to lead to goals. It is no simple matter to fully get at the sense in which situation x resembles or does not resemble situation y, and to recognize in this resemblance the likelihood of a particular number of goals being scored. However, the proof is in the pudding as it were, as xG models tend to perform impressively well at predicting teams’ future form; considerably moreso, for example, than simply looking at a given side’s form in the table and assuming that it will continue unabated along these lines. That xG has predictive value is demonstrable; less clear is the extent to which it can fairly ascribe a quality as vague as “greatness” to a side purely in terms of matches that have already been played.

In their arguments about the supposed greatness of particular sides, football fans have a certain charming way of horse-whipping whichever statistics they are able to get their hands on into supporting their beliefs or narrative understanding. The significance of particular out-of-context statistics is emphasized beyond all proportion while others are carefully cherry-picked and discarded until this team looks terribly overrated, that player is secretly the most important midfielder for their club side, and every club and international side in the world should sack their manager and immediately hire someone who will obviously improve them. This is the case for statistics as primitive as ball possession and shot count, and increasingly also with xG. It is for this reason that even the most vehement supporters of the application of statistical analysis to football are generally inclined to cringe when statistical arguments are brought up by television pundits, message board commenters, or anyone else seeking to bring their expertise to bear on the beautiful game. It is not particularly galling when someone who appears to grasp the function of and intentions behind xG suggests that if a certain side keeps on creating the same kinds of shots in the same volume, they will start to score more goals because those shots can be predicted to lead to goals at such and such rate; it is, however, quite a nuisance when fans of a club will promote xG totals that suggest that their preferred club should have rightfully won particular matches without hesitation, but when the shoe is on the other foot either become conveniently amnesiac about the existence of xG models or scorn them for being unable to capture the intangible ways in which their club is actually very good. It might be compared to the supporter who seems perpetually outraged about every penalty decision to go against their side but always willing to offer up a devil’s-advocate sort of a defence of those decisions that go in their club’s favour. The problem, fundamentally, is with the tendency of football supporters to use any and all alternative evidence as a consolation when they don’t have a result to celebrate, and to completely neglect the significance of evidence and argumentation in general when they do.

Obviously not every person who brings up xG in relation to a club’s performance is motivated by spite or bitterness at trying to rationalize away a bad result. Even entirely earnest supporters of a given club can recognize when their team has had a slice of luck in a match if they have any sort of emotional maturity, and by that same token it is not impossible to interpret xG dispassionately and use an understanding of the concept to make arguments that are not entirely “tribal” in nature. What is clear is that a team’s “underlying numbers” are not simply a proxy for their absolute quality or historical greatness - it would be incorrect to suggest that they are no more so than the actual results attained by a side, as a glance at “fluke-y” real-life title runs like Chelsea in the Champions League in 2012 or Greece at Euro 2004 demonstrate, but even the most dyed-in-the-wool zealots for data analytics will happily acknowledge that there is more on Heaven and Earth than is dreamt of in their philosophy when it comes to giving a truly thorough account of a side’s quality. Even in a thought experiment where xG has reached a state of perfect fidelity to the true quality of goal-scoring opportunities, it would only go so far in portraying the qualities of what footballing sides do on the pitch.

A decent analogy for the relationship between actual results and underlying numbers in football might be mind-body dualism. Critics of dualism will accuse the theory of painting a simplistic picture of a mind as an entity and the body as a separate entity and its seat in the material world. In reality, dualism as a theory posits agency-in-general as an order emergent from matter, with minds being specific agencies and bodies belonging to the order of matter. The relationship between the two is something to be parsed out from the sense in which complex dynamics within one conceptual order can supervene on another. It’s no small matter forming a coherent picture in this way, but also, it is of total importance to recognize that the mind and the body are not and can not be at odds with one another conceptually speaking, and that neither one is merely an illusory determinant of the other. Similarly, in an ontology of football, actual results would seem to supervene on underlying performances, and vice versa. As the mind is neither an entity which “takes control of” a body nor simply an epiphenomenal illusion determined by bodily function, an underlying performance in football is not determined by a final scoreline, nor does it directly cause the result on its own.

All things considered the relationship between underlying performances and results in football is probably a lot less conceptually difficult to grasp than the relationship between the body and the mind. Does the performance rule the scoreline or does the scoreline rules the performance? We don’t not know. The entire scope of the relationship of mutual and incestuous overdetermination between the two is for another, significantly longer essay, although a few examples of the form that the relationship takes might be helpful. Let’s say that a particular team.. Crystal P... no, that’s too obvious, let’s say.. C. Palace, take fourteen shots in a match, mostly from inside the penalty area, and score three goals while conceding a paltry three shots, all of them from outside of the box. They cruise to victory and knock League One Gillingham out of the FA Cup and are rewarded for their dominance of the ball, superior possession structure, and greater technique, pace, and physical strength as evinced by the run of play. Maybe in another example, Bologna find themselves in Serie A’s relegation zone in December after a run of results that analysts would characterize as unlucky, react by firing their manager, and their style of play is tangibly altered by his replacement; at times their new, more counter-attacking style is effective at procuring shock results against traditional “big sides” but their attacking production falls off a cliff in general and they dig themselves deeper into relegation territory. In one more example, we might imagine Borussia Dortmund starting a season very strongly by picking up sixteen points from eighteen, despite only managing to outshoot their opponents in three of their six matches, and having on average 46% ball possession. Eventually opponents start to see them as being particularly intimidating, and adjust their style of play accordingly; in contrast, Dortmund’s “confidence players” get a boost and the team begin to control games more and play more through-balls for their fast attackers to convert into scoring chances.

In each of the above scenarios, it is obviously the case that the underlying mechanics of how each team moves the ball around the pitch, defends without the ball, and attempts to create chances are at once designed with the sole or primary intention of creating more goals than are conceded, and not actually guaranteed to achieve that aim regardless of how apt they are to do so. The moment that a team actually starts playing a competitive fixture, there is a result, although it can be altered at any point up to the final whistle. The moment that there is a result there is a league table, or an elimination scenario at hand, or whatever, and these things prey on the minds of even the best players and managers. Footballing sides tend to play more conservatively when they have even a slender lead, perhaps to a fault at times. Factors like “momentum” and “confidence” are borderline illusory in actually determining tangible results, but it is possible to take even perfectly rational skepticism too far -  sometimes a side looks demoralized or sloppy in a way that is simply too palpable to be dismissed as nothing more than the projection of a pat narrative onto a more nuanced and chaotic reality.

The point of the above examples is simply to show that the senses in which underlying performances and results supervene on one another are numerous and take place in a world of messy complexity - football is, after all, highly messy and complex. Tactically and athletically strong performances often lead to a positive differential of high-quality shots, which in turn often lead to a victory expressed in goals scored, but this causal chain sometimes breaks down for reasons that are pretty clear and comprehensible (one goalkeeper is exceptionally good or exceptionally bad, an otherwise excellently-performing side gives away a cheap penalty or falls victim to a dubious penalty decision), and for reasons that aren’t particularly - sometimes one side simply can’t hit the broad side of a barn with their shooting despite getting into good positions, sometimes every shot seems to go in at the other end.

It is debatable, as well, whether a result “overperforming” an underlying performance could really be attributed to anything like luck if it is based for example on a spectacular performance by one side’s goalkeeper - the goalkeeper is as much a part of the team as any of their outfield teammates, and should they play exceptionally well and stymie a strong attacking performance by the opposition then it would not be any less accurate to call this a good defensive performance by the team than it would be in the case of one grounded in a similarly dominant performance by one or both of the team’s center-backs, or in a collective approach oriented towards quickly winning possession of the ball and then keeping it to prevent the opponent from doing anything dangerous with it. All of this gets at the whole conceptual issue of what exactly a team is as an entity, and of what role individual players as entities have in it (could they be best described as “composing” it? “Instantiating” it?). This scenario is particularly relevant to the question of how much xG models can render the “true” quality of underlying performances - xG will always tend to render handy shot-stopping performances by goalkeepers as defensive “overperformances” compared to the underlying defensive performances by given sides in given matches, but if the keeper in question is really that good then their innate goodness must surely be considered to be part of the broader goodness of the team rather than something outside of it, and the defensive performance really is a good one even if it happens to be inordinately concentrated in a set of actions which xG models tend to view as exceptional to the run of play rather than an intrinsic or holistic part of it. One could go on in this sort of regress.

None of this preamble is meant to weigh in on the suitability of xG to facilitating the many predictive and analytical functions that it does. For all the (endlessly improvable) flaws that every xG model has in how it xG’s, the concept is sound and is indeed a sophisticated and reasonably accurate way to represent the dynamics within football players, sides, matches, seasons, leagues, and so forth. The above question of how to account for quality of goalkeeping performance is not really a knock against xG, because its suitability to perform the functions that it performs hinges on it behaving the way that it does when faced with situations where a goalkeeper saves several very high-quality chances. What is far less clear, however, is how suitable xG is as a tool for arguing how truly good or great a footballing side is. What is clear is that it is not perfectly suitable, but we already knew that. Equally clear is that results alone also tend to distort and exaggerate certain facets and qualities of footballing sides are, and should not be considered fully reliable representations of them. The “actual” answer clearly lies somewhere between the two, though it shouldn’t simply be exactly half-way between. If we go back to the goalkeeping question, then it is clear that a model behaving in a particular way in order to do predictive analytics is not necessarily behaving in a way that is suitable to the purpose of supporting the arguments to be had about how “good” or “great” a side is - historically great sides almost always have great goalkeepers, and great goalkeepers should tend to overperform against the xG totals of the shots they face by definition. This is one clear instance in which xG is simply a bad way of trying to support claims about footballing greatness.

Nevertheless, while it is not on its own a perfect or even adequate way of rendering all of the things that would tend to confer historical greatness on the legacy of a given footballing side, and cannot be understood to represent the “absolute” quality displayed in any one match or over time, xG is a perfectly worthwhile tool to at least start to get at what a team really was beneath the numerical record and beneath the emotional impression left by “big” moments such as important wins or emphatic points advantages run up in the league table. The key is to not simply use a given side’s xG difference (rather than pure goal difference), or invent some kind of alternate league table based on the “xG results” of each match played in a season. Rather, an argument as to the greatness of a side one way or the other using xG should be based in interpretations of the xG totals that render an impression of how that side tended to play, and subsequent comparisons of that impression against those of other “historically great” teams. These interpretations should first and foremost treat xG not as infallible, but as a reasonably apt stand-in for having watched and carefully analyzed every single match that a team played - as such, they ought to assume that xG models render something that resembles the subjective perception of “footballing greatness”, and should aim to parse out in the xG data something that could just as well be parsed out subjectively in the matches represented by that data were one to have the time, inclination, and depth of understanding.

It is in this spirit that I will briefly look into Liverpool’s mind-blowing pre-coronavirus run of league fixtures using only the xG totals for the matches. Obviously I have watched quite a few of them and have opinions and arguments to be make about the quality and style of the team’s performances based on by qualitative, subjective impression of them, and I would be remiss to try and pretend that those feelings are not going to colour the broader points that I make or guide me in one direction or another along my way. However, I shall endeavour to do so anyway, having laid out in detail the conceptual caveats to my argument. The data I will be using comes from fivethirtyeight, although I could just as well use the xG totals created by the Understat or Caley Graphics models. I’m making a point not to look at figures for possession or anything like that - this is not an analysis of how they do what they do stylistically speaking, but only of how good the particular metric of xG believes (in whatever sense a model for aggregating data can believe or disbelieve anything) they are at doing it.

I will say pre-emptively that I do personally think that there are aspects of Liverpool’s style that are “unheralded” by existing xG data, which is to say that I think that there is a tactical sophistication and discipline to the way that the team behaves collectively in different temporal situations against different opponents and down to the level of specific situations like free-kicks or throw-ins in different areas of the pitch, determined by factors such as the personnel on the pitch, the opposition tactical setup, and the nuances of the game-state, and this fluid and dynamic sophistication is not easily represented in a temporally “flat” xG map. Obviously the incongruity between xG and results could also be put down to matters like finishing variance (debatable) and goalkeeping “overperformance” (Alisson is a great keeper!), but also it seems clear that as all teams naturally try to play with stylistic adjustments to suit the exigencies of particular in-game situations, Klopp’s Reds are basically as good as it gets at tailoring their approach to play as well as they need to when they need to without being gratuitous. Nevertheless, a team can only be so good at this sort of thing, and I would not want to fall into the fallacy that a side must be “making their own luck” whenever they get exactly the right goals at exactly the right time rather than just being smiled on by the football gods. The choice to use xG figures is kind of already a clue as to my intention to “bend the stick” away from Liverpool to some extent.

There are of course as many xG situations as there are football matches, and even if one were to “categorize” games by xG it would be entirely justifiable to create a fairly large number of categories and position them as meaningfully and saliently distinct from one another. It is in defiance of this spirit that I will be dividing xG results into precisely four categories. There are narrow wins, narrow losses, big wins, and big losses. Obviously there are a virtually infinite number of nuances to the distinctions between even fairly similar-appearing xG results, but if I wanted to get at those I would just write about the matches themselves. The use of the xG totals will necessarily be quite blunt and devoid of nuance because the point is a broad one about the extent to which Liverpool’s numbers are those of an “historically great” side - the nuances will hopefully come through in the interpretation and explication of the rates at which certain very broadly classified kinds of results appear, if they do at all. I will consider a total of 1.0 xG or more to constitute a “big margin” (so 0.5-2.9 is a “big loss” and 1.8-0.7 is a “big win”), and narrower margins such as 1.7-1.4 to be narrow results. Even in a case where one side looks pretty demonstrably superior - say, 2.0-1.2 - it’s not outside the realm of possibility that the “worse” side by xG will manage to win that match, and downright probable that the match will end a draw. As the margin gets wider in xG, the probability of course goes up that one side in particular will win the match, and if they somehow don’t that someone who watches the game in the stadium or on television, or even just peruses the xG or simply the topline statistics while checking the result, will wonder to themselves “how did they not win that?”

Of the four above-mentioned categories of xG results, which would one assume to characterize most matches played by historically great sides? Obviously historically great sides should very rarely suffer “big losses” in their underlying performances if they do at all. Are “narrow losses” permissible? Some, yes, although too many should surely render a side merely “quite good” rather than truly excellent. The matter of “big wins” and “narrow wins” is murkier - even the most dominant sides are going to have some underwhelming matches where they win narrowly or drop points when they “should” win or play dominantly on paper, and xG is going to reflect this. An additional problem with “narrow” wins is that, as mentioned above, they are inherently not guarantees of wins in reality, and in fact should guarantee a certain number of real-life draws and losses in large enough quantities. A “great” side should probably hope to “win” the xG by a margin of at least 1.0 more often than not to better ensure a steady diet of wins - playing a little bit better than the opposition all the time is playing with fire and when a team usually only shows the ability or inclination to create a few more good goal-scoring opportunities than the opposition put in front of them then that team is very likely to stutter up the table rather than rack up wins at the level of a peak Bayern, Juventus, or indeed peak Liverpool.

What, then, do Liverpool’s twenty-nine matches prior to the postponement of football due to the coronavirus look like? By fivethrityeight’s shot-based xG totals, Liverpool have generated at least 1.0 xG more than their opponent in sixteen matches, have generated less than 1.0 xG more than their opponent in eight matches, have generated less than 1.0 xG less than their opponent in four matches, and in one lone case, their inexplicable 0-3 debacle away to Watford, were out-xG’d by a margin greater than 1.0. This means that in more than half of their games, the Reds have shown a sufficient superiority that the xG rendered a “big” win for them. In only a small minority of their games did Liverpool fail to procure at least a narrow xG advantage, and when they did get outplayed by their opposition it was by an average of 0.3 xG if we exclude the Watford match. Given that the aim of the sport of football is to score more goals than the opponent, and the key process involved in doing so is to create more and better shots by hook or by crook, Liverpool this season have been a very strong and effective team. If one were to try and characterize this team as a mediocre one riding their luck, then one would be in contravention to the evidence of the xG, which has them winning at a canter more often than not, and otherwise tending to keep things tight enough that they give themselves every chance to take three points. There is not some easy case that “the xG proves” that Liverpool play pedestrian football and rely on lucky breaks to win routinely, and in fact a much clearer case to be made that they really have been a finely-tuned winning machine.

There is a difference, however, between being a finely-tuned winning machine and a record-breaking, historically great winning machine. It is much harder to make the case for Klopp’s team being the latter using the xG. Obviously any side outside of a “one-horse” league is going to rely on a little bit of luck and helpful variance to win twenty-seven games out of twenty-nine, but there certainly appears to be more than that in Liverpool’s twenty-nine game run. What is the likelihood of winning sixteen games in which the xG margin favours the team in question by 1.0 xG or more (bearing in mind that this is inclusive of some true blowouts of three or four entire xG)? It seems decent. Obviously even true juggernaut teams of the ilk of Bayern, Barcelona, or PSG have been known to drop points in games where they’ve kept the pressure on from first whistle to last and created a mountain of decent attempts while conceding almost none, and I personally would expect Liverpool to fall victim to this kind of funhouse-mirror 0-0 or 1-2 once or maybe twice if I were to look at the xG totals for all those sixteen matches without knowing what actually happened. Then again, it is not hard to look at them and think, yeah, sure, these are clearly dominant performances and sixteen wins from sixteen is entirely within the realm of possibility. It is worth noting also that these xG totals do not represent some streak of flat-track bullying - they are inclusive of wins against the likes of Wolves, Man United, Arsenal, and Sheffield United in addition to some mid-table and relegation-threatened fodder. Doing that really is impressive.

Also worth considering is that plenty of “lucky” wins achieved after the eightieth minute came in games with this sort of xG margin, including wins against Aston Villa, West Ham, and Norwich. The 1-0 win against Sheffield United in which United keeper Dean Henderson “gifted” the Reds an easy win with an error was, by xG, fully merited. Maybe most absurdly, October’s then-top-of-the-table clash against Leicester City, in which Liverpool won with virtually the last kick of the game after winning a cheap penalty, leading to plenty of hand-wringing as to the purported “luck” favouring Klopp’s side, finished with an xG margin of 4.4-0.1 in favour of Liverpool. In reality they were probably not, um, forty-four times better than the Foxes in that match, but if you strip away the dramatic circumstances of the late penalty it is pretty clear that they comprehensively outplayed one of the better teams in the league, and probably one of the twenty or so strongest sides in Europe; had Leicester walked out with a 1-1 draw it would have been something of a smash-and-grab.

The likelihood, however, of winning all but two of the remaining thirteen matches is where Liverpool really do seem to look like a team with a better points total than their underlying performances “deserve”. The Watford match is totally anomalous, and no amount of “luck” could keep that one from being a loss; the sole dropped points over the course of the rest of the Premier League campaign came in a 1-1 draw at Old Trafford. xG actually had the Reds as marginally better in that match, although it came out close enough that a sharing of the points seems just about right. Liverpool could even feel aggrieved given that Man United had a go-ahead goal stand when it really ought to have been overturned by VAR, although on the whole it was not some highway robbery - neither side were spectacular, it was a hard-fought draw and earning a point at Old Trafford is not something any Liverpool fan would ever shake a stick at. In the remaining games, we find Liverpool winning eleven games from eleven with an average xG margin of 0.31. This just frankly seems insane. No doubt some of this is down to Klopp’s side doing tactical things that enable them to eke points out of close games rather than being pure good fortune and variance - this is a pretty time-honoured strategy in the beautiful game of “fast kicking, low scoring, and ties! ties! ties!”. Still, that kind of totalizing domination of close matches is unprecedented and wildly improbable any way that you slice it, and even a notoriously “pragmatic” manager the likes of Marcello Lippi or Jose Mourinho would surely scoff and shout “it can’t be done!” if one were to propose to them trying to get that many points while keeping things that tight.

It is worth thinking about the manner in which Liverpool’s games were close, when they were. Nearly all of Liverpool’s “narrow wins” by xG had them creating between 1.5 and 2.2 xG while conceding a little more than 1.0. What can we very broadly infer about them from this? Even when this Liverpool side are not at their best, they are generally able to threaten the opposition goal to a respectable degree. When they are at their best they do so while also preventing the opposition from doing much in attack (and again, it needs to be emphasized, they are at their best more often than they are not in these twenty-nine matches). By contrast, when things get a bit dicey for them, they rely on their (really good) keeper or the profligacy of their opponents to keep them from conceding enough goals to nullify their hard work as an attacking force. This was not all down to the brilliance of the goalkeeper to the same extent as, say, Man United’s second-place finish in 2017-18 powered by David de Gea refusing to let anything get past him (bear in mind that Alisson was out injured for part of the season and his good-but-less-good lieutenant Adrian played in his stead). Then again, if this was the case then one could argue than the Brazillian is, if not Liverpool’s “best” or most talented player, their most “important” and “value-adding” player in the context of their unprecedented march to the title.

Liverpool’s four “narrow losses” by xG have been against Southampton, Chelsea, Man City, and Wolves; in all four games they came up against good opponents and were, if not genuinely outplayed, certainly stress-tested. These were close matches in every sense. Southampton, Chelsea, and Wolves could all cite slightly poor finishing or the simple matter of “sometimes they just don’t seem to want to go in” as to why they went home empty-handed despite keeping Liverpool’s goal at the center of the action far more than Jurgen Klopp would have wanted. City conceded three goals against a Liverpool performance which was brilliantly choreographed and genuinely ruthless to watch, but second-choice goalkeeper Claudio Bravo probably should have kept at least two of Liverpool’s goals out to make it a different topline story - it was not a “rampant” win by any measure other than the emphatic final scoreline. In all of these games, Liverpool did more than enough to give themselves a solid chance of winning, but to win all four they undeniably had to rely on the universe unfolding in several ways that it could just as or more easily have otherwise. There are other sides for whom one could cherry-pick four matches that they managed to win despite being narrow xG-losers, but these are sides who rack up far more narrow xG deficits than Liverpool, some of which inevitably result in draws and losses. To experience only four narrow losses in xG, and in reality win all four matches, a team does in fact require a good bit of luck on top of the quality of their underlying performances.

For the sake of comparison, it is worthwhile to look at the kinds of xG totals that other good Premier League sides have managed to get. Manchester United managed eleven “big wins”, seven “narrow wins”, eight “narrow losses”, and three “big losses”. In Chelsea’s case we find twelve “big wins”, nine “narrow wins”, five “narrow losses”, and two “big losses”. Leicester managed twelve “big wins”, five “narrow wins”, four “narrow losses”, and six “big losses” in addition to two games in which their xG total was equal to that of their opponent. In the case of Wolves we find seven “big wins”, thirteen “narrow wins”, seven “narrow losses”, and one “big loss” in addition to one xG “draw”. These are all very good sides, and when one expresses their respective seasons in the above way it is clear that they have all put up the underlying performances to justify playing European football next season. What is equally clear is that Liverpool have been a cut above all four sides, each of whom have won the Premier League in recent memory with the exception of Wolves. According to xG, Liverpool have thoroughly outclassed their opposition such that they generated an xG gap greater than 1.0 nearly as many times this season as any of the four sides mentioned managed to rack up an xG win of any description. Had a side with the quality of this season’s Chelsea or Leicester sides managed to win the league it would undoubtedly be a fluke of either an historically weak league or outright luck. And it is by contrast with the records of these entirely respectable footballing sides that Liverpool’s very real title-winning form comes into clear focus.

There is another good side in the Premier League, of course, whose xG totals have not yet been mentioned. Manchester City, the closest thing Liverpool have had to title rivals this season, managed nineteen “big wins”, six “narrow wins”, two “narrow losses”, one “draw”, and zero “big losses” in xG across twenty-eight matches (having played one fewer than Liverpool due to being involved in a League Cup final prior to the league being halted). Whereas Liverpool’s record would seem to show a highly strong team that doesn’t easily drop points, City’s looks like that of a team that plays against its opposition’s reserve side every week. Of course, the reality is that City managed a little over two points per game prior to the pandemic - this is nothing to scoff at, of course, and the fact that it is seen as a noteworthy underperformance is a testament to the quality of this City team since Pep Guardiola took over. Still, it does make the fact that City not only did not manage to win the title over Liverpool, but were barely able to make it a competitive race, something of a head-scratcher.

It is worth thinking about some of the matches in which City did drop points. The head-to-head match against Liverpool which effectively ended the title race as a contest was a truly even match that the Cityzens could hardly have felt aggrieved to lose - in fact, it very much mirrored the literally title-deciding match between the two last season in which City barely edged one of the most tactically interesting, precisely-executed, and balanced encounters that the Premier League has ever seen. Losses against Wolves and Man United (two each) point to City’s susceptibility to well-executed counter-attacking tactics, and even sides of less pace and quality than those have managed to break through the stifling press of Guardiola’s side and find themselves through on goal. It would be wrong to state categorically that xG models such as fivethirtyeight’s consistently underrate counter-attacking sides or overrate possession-based ones, but to an extent this seems like a reasonable explanation for the discrepancy between City’s impeccable xG record and their merely good real-life one. There is clearly something to the run of play in many of City’s matches that has facilitated a large gap between the frequency with which City take decent shots and the frequency with which their opponents do so while also causing the chances that City concede to be exceptionally damning ones. A similar phenomenon has chronically afflicted Atalanta in Serie A, whose accomplishments in the league table look incredible relative to their budget and historical stature but actually undersell the extent to which xG would seem to suggest they dominate all comers.

Positing that City’s at-times hapless defenders have not been up to the task of properly marking opposition forwards for the entire season, and that this has led to a landslide of scoring chances conceded the magnitude of which is not fully represented by the xG numbers is not an entirely baseless supposition. At the same time, there were certain matches in which the best explanation for City’s failure to win might in fact be some kind of curse. One could see how Norwich ran out winners in a truly open encounter early in the season, for example, but the likes of Crystal Palace and Newcastle United each snatched a 2-2 draw in matches that both the xG and the eyes and brain of anyone passingly familiar with the sport of football would suggest were one-sided affairs. Since whatever horrific spiritual crimes Pep Guardiola is evidently being punished for in the form of matches against Tottenham Hostpur Football Club are apparently beyond the scope even of last season’s Champions League quarter-final tie, City drew 2-2 in their first match against Spurs of the season and lost 0-2 in their second, from xG totals of 3.5-0.2 and 3.5-0.5 respectively. Two separate managers were beneficiaries of a divine moral law that had made the decision to use their club as an avatar of retribution against one bald Catalonian - there is as of yet no xG model that accounts for situations of this nature.

There are instances where xG “gets it wrong” which prompt analysts to look for ways that their models are not quite up to the task of representing the “real” dynamics of matches. On the other hand, there are matches that are much likelier to elicit a big shrug and an acknowledgement that sometimes the model is working fine, and it is reality which is stranger and more chaotic than the model is designed to reflect. Manchester City have straddled that line this season without question. In matches of the former case (say, City’s shock loss to Norwich City), we can start to see how the Cityzens are “not really as good as the xG has them”. In matches of the latter type (”NUFC 2-2 MCFC”) it appears to be more a matter of simply miserable fortune keeping City’s points total lower than it “should" be. When a side as well-oiled at getting and keeping possession of the ball and funnelling it into the opposition goal as City are get lucky they win a match 5-0 instead of 3-1, and when they get unlucky they lose a match despite “playing like winners”. This is an interesting point about xG models in general - they are meant to pick out trends within the chaos of football teams that win, lose, and draw matches fairly regularly, and are pretty well-suited to doing this, but aren’t designed or calibrated for sides that dominate every match that they play, for the good reasons that many more sides of the former type exist than the latter. When a team can only have bad luck, then any luck that they have will be bad.

If one were to accept that City lost some of their matches because they shot themselves in the foot by playing a high line and relying on inept defenders to clean up and not dive into challenges like imbeciles, but also grant that some of the above-mentioned results were freak occurrences and just give the points to City, they would be nine points closer to Liverpool, and so the Reds might well have something to think about going into the long, pandemic-forced pause in football. Still, they would be heavy favourites for the title and would be champions as long as their form was at all decent after the resumption of the league. Part of the issue here is that Liverpool simply have not been affected by the kind of rotten luck that City have. It seems simply a bit inexplicable that City could drop twelve points in matches where they clearly dominate the xG, and Liverpool end up winning the lot in matches with similar dynamics. If one were to obviate this difference by either bestowing City’s mysterious inability to turn these matches into three points on Liverpool, or making City as clinical in them as the Reds have been, there would still be a thirteen point gap between the sides, but things would seem less wacky. If City were to win their game in hand against Arsenal (spoiler alert from the league-resumed, fake-crowd-noises-added future, they have), and Ederson were to have started at Anfield and City nicked that game, there would suddenly be a four point gap and something resembling a title race. The margins are quite fine when one club isn’t under a cloud of weirdly horrendous finishing luck.

Pointing out that the margins in quality between City and Liverpool are in reality quite fine should not be taken as an insult to Liverpool - far from it, City are an excellent side and few predicted that they wouldn’t easily win the league at the beginning of the season. Clearly we have found a point where xG “fails to tell the whole story” about the dynamics in real-life football matches, as fivethirtyeight’s model portray City as a much more reliably dominant side than Liverpool and the league table tells almost the exactly inverse story - neither is entirely wrong, and neither entirely right. The Reds have been a greatly effective team who haven’t so much been without weaknesses as they have found ways to mitigate their weaknesses when they have had to. City have, for all of their woes, been an astonishing footballing side to watch on their day. It is very easy to see how Liverpool have gotten into the position they’re in, and similarly easy to see how City might have done the same if a small number of things had happened marginally differently. Head-to-head matches between the two sides in recent seasons have shown the two sides to be of roughly equal strength, and whatever the xG data has said, each has demonstrated in their own way the ability to amass large numbers of points and win the league. As City’s inconsistent league form mitigates their video game-like xG dominance, a look into Liverpool’s xG for the season does admittedly take some of the sheen off of their truly remarkable twenty-nine game run of dominance.

It is often said that every club that wins a title gets some luck somewhere along the way. This is not entirely true, as some simply have such a huge built-in advantage before a ball is kicked the even a run of misfortune isn’t enough to stop them from becoming champions - see Bayern or PSG for examples of this kind of situation. There is nothing wrong with celebrating a trophy that is the result of some kind bounces here and there or heaping adulation on the teams who win them, because even for the luckiest teams, luck is only the icing on the cake. Leicester City fans are always going to have their (perhaps a touch fortunate) once-in-several-lifetimes league triumph to look back on, whereas Bayern supporters are not likely to enjoy their umpteenth Meisterschale even more because a stats website suggests that they should have won it by an even greater margin than they did. What kind of a deranged person cares about such things? Liverpool haven’t been “handed” their first league title in thirty years by a corrupt FA through the introduction of VAR, nor is their title win a “fluke” in any sense. This team has been impeccable every step of the way (or every step not taken at Vicarage Road, anyway), and they’ve earned this. Klopp has earned this. The fans, God knows, have waited long enough for this.

As to the question of this team’s enduring legacy, well, that’s a bit of a different matter. Are they the greatest Premier League side of all time. No, probably not. Even accounting for their “luck” this is just a flat-out better side than, say, the 2011-12 Man City side, or Chelsea in 2014-15 or 2016-17. How about the Chelsea side of 2004-05, which finished with 95 points, lost once (0-1 to City from a Nicolas Anelka penalty), and conceded fifteen goals over the course of a 38-game season. That is a genuinely tough one, especially given that there is much less underlying data about sides from even fifteen years ago, and football tactics were actually significantly different than they are today. But in terms of some abstract “greatness”, no, it wouldn’t be obvious either way. That this can even be said of Klopp’s Reds is again no slight; that Chelsea team were immense, and the fact that a look through the xG of Liverpool’s pre-coronavirus season yields a profile of a team at that level of dominance does indeed put them in the bracket of an historical side of some description. They are, I would estimate, a better team than the Arsenal Invincibles, and better than any United side that I can recall watching. Outside of England, comparisons to Heynckes’s and Guardiola’s Bayern teams, Conte’s and Allegri’s Juve teams, Klopp’s Dortmund, and Real Madrid under Mourinho, Ancelotti, and Zidane are all worth making, and could in most cases be made favourably towards the Reds, though they could always go either way. They’re still nowhere near Guardiola’s Barca, that team is sacrosanct. But the fact that all of this can be written without seeming ridiculous or detached from reality is a huge compliment to this team’s accomplishments.

To finally really throw off the mask and start to editorialize: no, Liverpool in 2019-20 are not the greatest side the Premier League has ever seen - that would still have to be Man City in the two seasons past. The City of 2017-18 was not unlike the current team under Guardiola, and somehow had even more eye-popping numbers, but there were key differences that allowed them to actualize their xG dominance. The presence of German winger Leroy Sane seemed a crucial element in deciding the kinds of games that this season have tended to result in frustrating dropped points for City - the Cityzens’ attack has been at times too narrow this season and there have been occasions where they were crying out, as it were, for the injured Sane to break the cycle of sterile possession. The margins for teams at the level of City and Liverpool are fine, and the presence or absence of a player with the capability to take defenders on at the level that Sane is able to can make all the difference, it would seem. Also, Guardiola’s team have managed to concede both more goals and more xG in twenty-nine games this season than they did across the entirety of either of the last two seasons. Exactly why this defensive decline has taken place is a matter for another piece entirely, and there is plenty of good analysis out there probing tactical and personnel issues that have led to City having a soft center in a way that they did not in years prior. At any rate one should not let the flaws on display this season distract from how outrageously good City have been in the very recent past - those seasons are in fact harder to write about in some ways simply because the level of dominance got to be almost monotonous. No team is entirely flawless, and that side were lucky not to lose in the group stage of the Champions League to a Napoli side who frequently outplayed them, before deservedly exiting the competition to, who else, eventual beaten finalists Liverpool. But in the league, a three-figure points total in no way overstates the extent to which they were in another footballing universe from the rest of the competition, and they are still the one side who genuinely hit a qualitative level that no other side has in the competition’s history.

Whether or not Liverpool actually equal or improve upon City’s single-season points record will be of relatively marginal consequence in celebrating their title victory when they finally rubber-stamp things. Title wins aren’t academic exercises. They are on course for a century of Premier League points if they play at a good pace when football resumes, though it wouldn’t be a surprise if Klopp prioritizes keeping his key players well-rested for next season and introducing youngsters into the side over committing his team to the pursuit of the symbolic points record once his team are mathematical champions. Whatever the number ends up being, it should probably not be a major factor in how Liverpool are placed in a historical context, though it seems likely that an inordinate level of importance will always hinge on it, in the same sense that the symbolism of going a full season unbeaten will always have an imaginary significance attached to it that is out of proportion to the extent that it actually reflects the underlying quality of a side. Given the ersatz rivalry that has emerged between Liverpool and City, many Scousers will feel that a points total above 100 specifically proves that these Reds are better than City were when they achieved that total; this may be a persuasive argument along the lines of “football tribalism”, but in reality the historical record is likely to favour City’s title-winning sides as the more comprehensively good footballing side for reasons that are not identical with, but bear some resemblance to, the reasons that underlying performance data favours them.

A comparable historical example, since one of the broader purposes of this entire piece is to try and place Liverpool’s imminent triumph in historical context, might be Real Madrid becoming the first side to reach a hundred points in La Liga in 2011-12, as well as breaking the single-season goals-scored record. Jose Mourinho and Los Blancos were rightly lauded for the achievement, but apart from Madird propaganda arms such as Marca few truly believed that they had eclipsed Barcelona regardless of what the points total said. The next season Barca became league centurions as well, despite a perceptible decline since the glory days between 2008 and 2011. The next season Atletico Madrid won the title, and the season after that Barca signed Luis Suarez and won the treble not while playing the best football anyone has ever seen, but very possibly with the most cartoonishly good starting XI. The point is that the historical memory does still favour points totals, points-per-game figures, and so forth to some degree, but these are of less importance than the impression left by the actual football played by a side.

This Liverpool side will probably stand out more than the above-mentioned Madrid side in most fans’ minds, at least English-speaking ones, because of the catharic nature of their win for fans, the sheer oddness of the fragmented season, and the fact that no one could come close to reaching them in the table after about October. As recency bias fades, however, it seems very probable that they will nestle in and around that side among the imaginary hierarchy of great footballing sides. If anything the 2011-12 Madrid side is now underrated in the sense that the sheer level of their dominance is not always properly recollected outside the Spanish capital, and this fate seems unlikely to befall Liverpool unless they experience a run of success comparable to the one that Los Blancos did in the decade or so after that season that might make this remarkable season look like only the beginning. I am guessing that most Liverpool fans would be fine with this particular team becoming historically underrated if the reason for it were that it ends up getting obscured by an era of constant success experienced afterwards.

Sometimes a club gets relegated despite everyone seeming to agree that they aren’t that bad. It gets to be commonly acknowledged that they were unlucky. By that same token clubs make it into the Champions League places, or even win the league without necessarily being the best club in the country, and there is usually a kind of strange reluctance to call them even a bit fortunate. Did Chelsea get some kind bounces along the way to 93 points in 2016-17? Did the Leicester fairytale season involve a slightly improbable number of coin-flip matches going the way of Claudio Ranieri’s team? Of course. Were both of those sides the best in the league during their respective title-winning seasons? That’s up for some debate, but given that the teams who got closest to them in the table were Arsenal and Spurs, yeah, they probably were. Liverpool are in a similar situation. Are they lucky to be runaway league leaders? No, certainly not. Is there a decent amount of good fortune within their 2.83 points-per-game total? With equal certainty we can say that there is. The question of whether Liverpool have been lucky should not qualify any feelings of euphoria by the club and their supporters, or add any asterisks to the congratulations that they are owed. Nevertheless, the question of luck will surely come up as Liverpool’s current season becomes a matter of history and memory, and the question becomes not a competitive one of whether or not to award them the title, but a sports-fan-argument or football-publication-nostalgia-piece one of whether they were among the x number of greatest sides in living memory, or the best side ever to grace the Premier League. As to those questions, we should aver that yeah, Liverpool have been a bit lucky.