We have a model at StatsBomb Services we call Passing Ability. You've probably seen the percentile numbers in newer radar charts, and if you want more on the detail work, you'll find it here. It was created by the very talented DOCTOR Marek Kwiatkowski. We use this model in place of passing percentages because as I explain in the link, passing percentages actually tell you very little about whether someone is a good passer. We think this is better, and we use it professionally every day. Anyway, many of you probably already believe Lionel Messi is a great passer. His skill in this area is pretty easy to pick up just from the eye test, and from a stats perspective, if you had to say one area that Messi is quantifiably better than CR7, you'd pick his passing. However, when you plug the info into an expected pass completion model, Messi's passing takes on an entirely new dimension. Below is a visualisation we produced to help uncover a little more information about each player's passing and how it's graded by the model. The top set of bars is volume of passes per 90 in each expected passing percentage (xP) bucket. So the bucket to the far right are passes the model expects to be completed between 95 and 100 percent of the time. The grey line is the league average volume for players in his position. The second set of bars details whether they are performing better or worse than the model expects in each bucket. Tom Huddlestone is in the 50.9th percentile of all passers in the data set, and as you can see, his completion by bucket ebbs and flows. It's not a perfect model, but in the vast majority of cases it passes the sense test. It thinks Xavi and Toni Kroos and Mesut Ozil and David Silva are awesome and so does everyone else. Which brings us back to Messi... Messi is awesome in a way almost no one else ever has been or ever will be. The next chart is from Messi's passing output in 2014-15. Now what you have to remember about Messi is that he's marked more than any other player. He's subject to more double and triple teams around the box. He's bodied constantly. And his teammates are also often marked more heavily than other players in the model, because teams always collapse the box against Barcelona. The environment he's completing passes in is tougher than what's faced by almost anyone else. And the model doesn't KNOW... So the model is out there thinking, "Yeah, this guy is pretty awesome. And not only is he awesome, but he has an insanely high volume of passes for a forward in almost every difficulty range, which is even more impressive. But I have to treat him like all the other students because I am a lowly event data model and can't tell the pressure he's under for every pass... here you go: 99.89 percentile." But in reality, the degree of difficulty is closer to whatever the Chinese divers always do to obliterate everyone else at the Olympics, and Messi pulls it off just as easily. This season has arguably been Barcelona's weakest in terms of available weapons. Suarez has struggled. Dembele is injured. Neymar is gone. So it's very much The Lionel Messi Show, and more so than it's ever been. How has Messi's passing graded out this season? Just when you think you have investigated all of the ways Messi is ridiculous, he finds another way to blow your mind. This Was a Bit of a Xhaka One of the things that's been weird about Arsenal's start to the season isn't the results - the squad is quite good - but Granit Xhaka's drop off in pass completion. He's still completing 83% of his passes, but that's down from nearly 90% last season. Volume is also up slightly to 80 a game vs 72 in league play last year, and Arsenal are winning, so who cares? This doesn't have to be a big deal. Xhaka could simply be attempting more difficult passes regularly and his passing percentage would go down. I kind of shrugged my shoulders until I pulled up the info from the passing vis and it was unhappy as well. This is probably a blip. It's early in the season, the volume of passes isn't that high, and maybe teams are playing Xhaka differently on the ball. But it's also something to pay attention to going forward, because this year is different than every other Xhaka season I have data on. As an analyst in the team, I'd take a close look at this and see if I could figure out why this has happened, because maybe there's a tactical element to it we can solve going forward. Or maybe this is a happy trade-off for a style of play that is bringing us more success, and you're fine with it. Sometimes potential problems are simply something you want to know more about before dismissing. A Note About Imperfections The passing model does not like Ousmane Dembele. It grades him quite low, despite the fact that he's one of the best creative wide players in football. The vis I've been using in this article was built to help explain why this big ass model that uses 20 million open play passes is unimpressed. As you can see above, it thinks Dembele doesn't complete nearly enough of the "easy" passes. However, it also flags the fact that he's pretty good at completing the hard ones, which are exactly the types of passes that create goals for his teammates. Now there are options available to "fix" this problem. We could skew the weight of the rating to account for high risk/high reward passers to help boost their numbers. Or, we could scrap the model and start over, attempting to account for this problem, but potentially open up other issues in the process. OR, since we've learned a bit about where the holes are in this current model and we like how it performs almost everywhere else, we can just account for this with human judgement layered over the top of the data algorithm. Using data in sport is always challenging. A bit like being an athlete on the pitch, nothing you ever do will be perfect. You are typically faced with a series of sub-optimal choices, and have to choose the one that is least wrong. There are valid critiques of almost everything we do, but my response these days is often, "I agree with you, but show me something better." This is exactly why you can't let perfection get in the way of progress, because in a sport with the complexity of football, perfection does not exist. Except maybe in the case of Lionel Messi. Ted Knutson email@example.com @mixedknuts
Month: November 2017
Manchester City: Powered Up Under Pep
This article was co-authored by Euan Dewar (@EuanDewar) and Mark Thompson (@EveryTeam_Mark ). Manchester City. Doing alright for themselves aren’t they? It’s been a while since an English football team has loomed over the rest of Europe, hanging ominously in the air. Extraterrestrial. Imperial. “That’s no moon. It’s a Pep Guardiola team.” Through 12 matches the boys in blue sit atop the league with 40 goals scored (13 more than the team in second - their neighbours United). Four teams in last season’s Premier League didn’t even reach 40 goals after 38 matches. Their expected goals tally isn’t that far off either: +31.9 in terms of non-penalty xG and +34 overall. They’re blowing away all who stand in their path. Perhaps, because a certain strain of English fan has always been sceptical of Guardiola’s success, the dominance of this current City team isn’t yet fully appreciated. Not only do they reign supreme in the surface numbers, but how they go about the game doesn’t let their opponents have a say either. City average over five sequences of 20 or more consecutive passes per game, which is freakin’ phenomenal. The next nearest side in (Opta’s) living memory is the Barcelona side of 2012/13 with 3.9, and then Bayern of 2015/16 with just over three. Guardiola's influence here is clear. There are only two other sides who average more than 2.5 such sequences per game. City are doubling that. They also get to the final third in over 55% of their possession sequences, at the top of the historical list along with Pep's Bayern Munich sides of 2013-2016. For any who were sceptical about whether Guardiola could ‘do it in the Prem’, the answer is now clear. Patient build-up is only half the story with City. A defining characteristic of Pep’s sides is how they shift through the gears, moving the ball (and consequently the opposition) from side-to-side until they spot an opening, at which point they up the pace and cut through the defence. These facets to their game can be seen in the varied speed of their output. In the aggregate they’re as slow as you might expect, but look solely at possessions leading to shots and the threat is apparent. When the time is right, they go for it. Shots from counter-attacking opportunities represent a lower proportion of City’s attacking diet compared to the rest of the league, as you might expect from a team who dominate possession like they do. However, they still get plenty of them - the 3rd most in the league from transitions that last 25 seconds or fewer (with a transition defined as a possession that starts in a team's own defensive third and ends in the opposition third). This is also an even higher proportion than in Pep’s last seasons at Bayern. City’s transition possessions go through the right half-space at the highest rate in the league and left half-space at the second highest. This makes sense considering how much David Silva and Kevin De Bruyne love pulling the strings from these zones. When you combine that duo with the attacking menace of the likes of Leroy Sané and Raheem Sterling you get downright lethal counters like their second goal against Leicester at the weekend. The foxes went from attacking to picking the ball out of their own net within 15 seconds. The Other Side of the Ball Possession is nine-tenths of the law in some places, a criminal offence in others, but for Guardiola it’s also a way of defending. Not only can the opposition (obviously) not score if you’re the ones with the ball, but Pep also uses it to help his teams get into position to counterpress. City rank a clear 1st in the PL in terms of shots in the opposition box that come within 10 seconds of a turnover in the final third. As a result of this combination of defensive techniques, City’s opponents’ only get to the final third in 34% of their possession sequences. This isn’t quite at the historically good levels as City’s passing stats, but the closest rate in the Premier League this season is Liverpool on 39%. City allow their opponents just under 2.5 sequences of ten or more consecutive passes per game, far below the average of league average of eight. The next lowest figure is for Spurs, allowing just 3.7 per game, but there’s daylight in that distance between City and the London side. However, every strategy in football has a weak spot and City’s is the same as it always has been for teams in their mould. Dominating the play, pinning the opponent back in their half, it leaves space open behind that can be exploited. The thing is though, Pep has been coaching in a variation of this style for a decade now. He’s obviously aware of this issue and has spent plenty of time tinkering with the approach in order to minimize it. As a result, the opportunities the opposition gets are few and far between. While City give up a fair few transition opportunities, this season they have allowed the fewest that reach their defensive third as well as the fewest shots in transition. However if you can get into their danger-zones then all is not lost. City's underbelly isn't exactly soft by any means, but it isn't quite as suffocating relative to everything else. Really though, Manchester City aren’t the Death Star. The Death Star had a clear flaw that was easily exploited by a space farm-boy. City are scarier than that. None of this is totally permanent of course. Football’s a funny, volatile sport and things could get tough when the fixtures start to pile up. As things stand though, they look well set to impact both the domestic and European scene in a way an English club hasn’t for some time. __________________ Thanks for reading. Have yourselves a good week out there!
Marcelo Bielsa's House of Horrors
Lille were (and to some extent still are) one of the great experiments in European football. You wouldn’t think that a club currently residing in 19th place in Ligue 1 with only 9 points from 11 games could be classified as such, but ahead of the season, with new ownership and the return of everyone’s favorite crazy managerial genius in Marcelo Bielsa, it was hard not to be excited about which direction the club was moving. The club won a ligue 1 title as recently as 2011 but in recent years had been selling their best players just to balance the books. The takeover offered a chance for a fresh restart. Add to that an intriguing young roster bolstered by a £50m summer spending spree and it wasn’t unreasonable to think that in a league that was fairly chaotic outside PSG, Lille had a great chance to hit the ground running, sneak in and become something of a hipster’s favorite. Instead, the early excitement has been vanquished and the kind of depressing football that Lille fans came to expect with the previous regime has recurred. Lille have been very bad defensively through 11 games, and have shown a sterile attack that looks like nothing from Bielsa's last Ligue 1 tenure at Marseille during 2014-15. This was supposed to be the year where major progress occurred, perhaps leading up to the point where the club could viably contend for a Champions League spot in future seasons. Instead, we’re at the point where it’s fair to wonder if they’ll even be able to drag themselves up to a mid-table position at season’s end. Underperforming Attack For stretches of the season, in 2014-15, Bielsa's Marseille squad were the toast of European football (like a lesser version of what Napoli are today). It was the Bielsa concept at its finest: breaking up opposition as far up the field as possible to create good shots taken in transition, scoring a lot, and a major impetus to attack whenever possible. The physical demands that his tactics required led to the team's eventual demise as they tired through April and May, but it can’t be denied that during that season Marseille matches were appointment viewing television. This Lille side has been almost the opposite. The raw numbers are bad: Metz and Caen are the only two Ligue 1 teams to have scored fewer non-penalty goals, and only Amiens have created less big chances. Lille rank near the bottom in expected goals for. At an xG/shot rate of 7.5% in open play is super low and watching Lille settle for hopeless long shots instead of trying to look for better options has been grim.
(chart via StatsBomb Services)
Verticality is a major principle that Bielsa adheres to, the idea of getting the ball forward in a timely fashion and getting past defensive lines. When it’s truly flowing, it can produce scintillating football. Lille haven’t really been able to do that. A lot of times, the ball will be passed from one center-back to another then followed by a hopeful long ball to one of the wide players. In Lille’s first ten games, no team in Ligue 1 had a higher proportion of passes occurring in their own half of the pitch.
(chart via Euan Dewar)
On top of that, when they have been able to connect on passes to their attacking wingers, it’s not led to anything worthwhile. Luis Araujo and Anwar El Ghazi are talented wingers and their acquisitions at the time made sense, but they’ve been settling for bad shots far too many times at the expense of potentially better opportunities. They've had fullbacks making runs to advantageous areas to make crosses or cut-back passes, but instead the duo have taken bad shot after bad shot: https://streamable.com/ax8v5 There's no guarantee that passing to Kevin Malcuit or Fode Ballo-Toure whenever they get into these situations will net a great chance, just as there's no guarantee that recycling the ball in the final third will create the vital opening needed. But settling for these kind of shots time and time again is a form of malpractice. Perhaps the greatest symbol of disappointment comes from the form of Yassine Benzia, a once highly rated attacker who even was being showered with praise from Bielsa months ago. It’s quite hard to have an attacking player who offers next to nothing as either a goal scoring threat or as a playmaker but Benzia is managing to pull that trick off. From his time at Lyon forward, he has been hyped as a prospect, but we have very little evidence so far into his career that he’s a good player, and it’s even more pronounced when he’s playing as a #10. He’s just not been dynamic whenever in possession of the ball. His touch has lead him astray on numerous occasions, turning intriguing possibilities into distant memories. Despite the fact that he’s creating 2.7 key passes per 90 minutes, almost all have ended up in long range shots. Not all of that is his fault, but the experiment of him playing that position has not been a success. The decision making of this squad whenever they do get into good positions has left a lot to be desired and to some extent, it’s expected. Benzia is still 23 and hasn’t played a lot of Ligue 1 football, Araujo is 21, El Ghazi is 22, Thiago Maia is 20. I’m not sure that youthfulness can be used to fully explain a team that’s shooting below 8% in open play, but I’m willing to believe that it’s can be a trying exercise to coach a young squad to take greater care of the ball and value shot selection, if indeed they are being directed that way. In a season full of disappointment, one of the few rays of sunshine has been Thiago Mendes’ play. Being a central midfielder in a Bielsa led side isn’t the easiest of tasks, but he’s been productive. His ability to make himself available for passes has been a welcomed addition, and he’s flashed the ability to be a press resistant midfielder by being hard to knock off the ball. https://streamable.com/0f556 Are there hopes for Lille turning things around in attack? Against Metz two weeks ago, they were able to string together passes but even then, it never led to any high-quality chances in open play. There was still the repeated ball circulation between the center backs leading to inaccurate long passes towards the wing, with a lack of occupation in the middle portion of the final third continuing with a focus on wing play. The 3-0 scoreline was flattering, and the match itself didn't feel like an indicator for a possible turnaround in fortune. A man-marking press exposed There are different ways to defend as a team, but an obvious main goal is to allow as few quality chances as possible. There are teams like Liverpool who press and do fine in terms of shot volume but the chances they do concede tend to be memorable and dangerous. Arsenal have generally done well in conceding long range shots despite employing a half-hearted approach to pressing opponents. Chelsea try to disrupt things whenever opponents get into the middle third. We know the deal that comes with Bielsa coached teams; a holistic man-marking approach that has become somewhat of a rarity in European football. It’s designed to break up buildup play and attack as quickly as possible. To some extent, you can see how it’s worked out with Lille. They have been able to break up opponents passing very high up the pitch and disrupt build-up play. It hasn’t necessarily led to them creating quality chances, but they’ve regained possession of the ball higher up more times than not.
(chart via Euan Dewar)
With all pressing systems, you’re at the opponents’ mercy if bypassed. Lille have tested that because once things get broken, it’s led to some chaotic situations. Teams coached by Bielsa are inherently dealing with a certain level of chaos, and this season it hasn't ended up to the benefit of the team. Whenever they lose the ball, players frantically try to find someone to mark with the demands the system entails. The result has been moments where the opposition can carve them up and get into dangerous areas. https://streamable.com/3n9lu Lille are towards the bottom in both shot volume and expected goals conceded, and it's stuff like this along with some weird individual defensive errors that has been the cause of that. I'm probably more confident that there will be some form of improvement in this area compared to the attack. Even though the end results have yet to gel, there are things that can be pointed at as signs of the players learning to carry out what is expected of them. In attack, you'd be hard pressed to find any positives. Will Lille get Relegated? Looking at the state of affairs within Ligue 1, and on a pure talent basis: Lille shouldn’t be in the position that they are in. People (like myself) overrated both the quality that Bielsa was working with when the season began, and just how hard it could be to coach up a team that’s full of players aged 23 and under. But even the biggest of pessimists couldn’t have imagined just how bad it’s been. There are further intriguing players on the squad like Nicolas Pepe and Fares Bahlouli. El Ghazi is a better player than he’s shown to be so far, and the defensive personnel in theory is as good if not better than the unit Bielsa had to work with at Marseille. On the other hand, this team has been very bad through 11 games and if they continue to play in this manner, they will stand a great chance of dropping into Ligue 2. Lille can’t buy a good shot on goal, and despite the number of athletes at his disposal to use for the trademark press, the scheme has been leaking chance after chance once the press gets broken. Losing a player like Nicolas de Preville at the end of the summer window has proven to be a bigger blow than might have been thought, a productive attacker that could've been leading the line alongside for Araujo and El Ghazi. And if the two of them continue to be high usage wingers who provide little to no value in attack, then it might very well be that the team as a whole continue to create shots at an 8% xG/shot clip throughout the rest of the season. Put a gun to my head and I still think Lille are going to dig themselves out of this hole. But the top 6 prediction that I had for this club seems almost laughable considering what has occurred on the pitch. Lille have been bad and with little signs of improvement through November, Bielsa's European return has turned into something of a nightmare.