Month: November 2018

It is difficult to believe that at some point over the next five or six years, Paris Saint-Germain won’t look back and regret their decision to provide Giovani Lo Celso with an accessible pathway out of the Parc des Princes. Lo Celso arrived at the club in January 2017 after making his name in his native Argentina as part of the attractive Rosario Central side managed by Eduardo Coudet that reached the quarterfinals of the Copa Libertadores in 2016. Playing as an attacking midfielder, he not only provided goals, assists and incision but at least one “wow” moment every match: a delicate touch, a scooped pass, a confounding flourish. Playing time was severely limited in his first six months in Paris, but he worked hard to settle in and prove himself useful to coach Unai Emery. He was rewarded with a good set of minutes last season, making 19 starts and 21 substitute appearances across league and Champions League play. He made up part of the midfield three, and was even sometimes employed as the central defensive midfielder - a role he had never previously occupied. Given those demands, he put together a very solid campaign. The low point of his season came in the first leg of the club’s last-16 exit from the Champions League at the hands of Real Madrid. he was rather hung out to dry in that central role in front of the defense (the lack of alternative options perhaps demonstrating some of the same lack of foresight that led to PSG loaning Lo Celso out this season), his inexperience showed. He struggled to establish any sort of presence in the match and conceded a penalty in a 1-3 defeat. It was a blot on an otherwise promising campaign, one in which Lo Celso displayed the range of his abilities. Fully settled, he seemed set to push on. However, just 10 minutes of action over the course of the club’s first three matches of the new season suggested he wasn’t to the taste of new coach Thomas Tuchel. On the last day of the summer transfer window, he joined Real Betis on an initial one-season loan, with a €25 million purchase option (mandatory if Betis qualify for Europe; executable without PSG’s further approval if not). It already looks an excellent match. At a top-line level, he has provided five goals and two assists in league and Europa League action, at a rate of a direct goal contribution of 0.72 per 90 minutes. He stole the show in a standout performance in the thrilling 4-3 win away to Barcelona earlier this month, impressing with his ability to make space for himself with good touch and awareness of his surroundings when receiving - attributes also evident in his ultimately futile late strike away to Villarreal on Sunday. It is difficult to decide which radar best encompasses Lo Celso’s output to date. Betis usually line up in either a 3-5-2 or 3-4-2-1 formation. In the former, he is part of midfield three; in the latter, one of the two attacking midfielders behind the central striker. Wherever he is positioned, however his role is defined, his numbers are impressive.   In attack, he completes more dribbles and advances the ball into the final third more often than any other Betis player, while also ranking in the top three in terms of xG and xG assisted. In defense, he leads the team in tackles and pressures, and ranks second for pressure regains. More broadly, no other player in the big five leagues this season with 500 or more minutes of action under their belt has been able to match Lo Celso’s combination of an xG contribution (xG + xG assisted) per 90 of 0.30 or higher, three or more successful dribbles, nine or more deep progressions, 15 or more pressures and over 3.5 tackles and interceptions (indeed, on a possession-adjusted basis, he is making 5.04 successful tackles per 90, seventh best in the big-five leagues). In short, he is providing a uniquely well-rounded contribution. Unsurprisingly, Betis have made it clear that they want to take up their purchase option next summer. They view Lo Celso as a player around whom an ambitious project can be built. He, too, has suggested he is keen to stay in the south of Spain. If he continues to perform at his current level, market forces may determine otherwise, but at this time, the partnership between Betis and Lo Celso is clearly a mutually beneficial one. If his high number of turnovers (4.73 per 90, sixth highest in La Liga) means he still cannot quite escape the assessment from Brazil’s 1970 World Cup winner turned columnist Tostão that he is a player of “many moves, few decisive,” he has, in Quique Setién, a coach committed to providing his most talented players with the necessary supporting structure to take the risks required to provide incision. To provide moments of quality like the back-heel that released Loren Morón into the area to score against Celta Vigo. Given his capabilities, it remains strange that Lo Celso was one of two outfield players not to receive a single minute of playing time for Argentina at the World Cup. Jorge Sampaoli cycled through almost every other combination of personnel and formation available to him but never once called on Lo Celso. Meanwhile, Lionel Messi was left having to do pretty much everything by himself in attack: progressing the ball, setting up chances, taking shots. There were suggestions from inside sources that the coaching staff were concerned about Lo Celso’s ability to handle the pressure of representing his country at the World Cup, but current interim (possibly soon permanent) head coach Lionel Scaloni was part of that group and has still made the 22-year-old the centerpiece of subsequent national team friendlies. Lo Celso therefore looks to have bright future ahead of him at both club and international level. The relatively meager figure of €25 million that PSG placed on his head already looks a startling misjudgment on their part. Betis have a player they can enjoy and derive success from right now, and a prime asset if and when he outgrows them.   Header image courtesy of the Press Association

Is Lionel Messi slowing down?

This is a sacrilegious thing to suggest. Messi is a god. Crucially though, unlike the whatever higher power you choose to worship, Lionel Messi is also aging. He’s 31 years old. That’s a good three years past when most players, even the best ones, begin to lose their fastball. There’d been precious little evidence of that with Messi though. He’d evolved his game, morphed from the greatest creative goal scorer on the planet to the greatest creator on the plant….who also happened to be a thirty goal scorer. What Messi was doing on the pitch changed, but his production was as otherworldly as ever.

Before we take this dive, the first thing to make clear is that Messi is still unreal. Compared to people not named Lionel Messi he’s still just better than them. The question is not, how does Lionel Messi compare against the other mere mortals he deigns to take the pitch beside? Rather, the question is, how does Messi compare against past versions of himself. And, there’s reason to believe that from last season to this there’s been a slight decline.

The first, and most obvious place to look is the scoring numbers. Last year he was a major goal scoring force. He scored 34 goals, 32 of them from open play. That was way beyond his 21.97 open play expected goals. But even his expected goals were impressive. His 0.63 expected goals per 90 were the third highest in the league, behind only Cristiano Ronaldo and teammate Luis Suarez, an especially impressive total since they were accompanied by 0.38 expected assists per 90 minutes by far the most of any player with over 1000 minutes played.

This season, so far, his goal scoring has taken a noticeable hit. His 0.48 xG per 90 is still very good but it’s a decidedly new occurrence to refer to Messi’s output as merely very good. He’s the sixth best scorer by xG per 90 in Spain and taking a step back and looking across Europe’s big five leagues. He’s not in the top 10, or even the top 20. He’s 30^th in Europe by xG per90 (among players with over 500 minutes played). That’s…well, it’s certainly different for Messi. And it’s easy to see from his shot chart that he’s simply not getting great shots this season. Most of his shooting is coming from outside the box, and even those shots which are from closer to pointblank range still aren’t the kind of clean red hot looks Messi typically ends up getting on the end of.

 

 

None of this would be at all concerning if Messi’s slowly declining scoring rate was being offset by a rise in his creative numbers. That’s what’s happened over the last couple of seasons. Messi has become the best attacking creator in the world. Except that hasn’t really happened.

Messi’s expected assists numbers have nudged up slightly from 0.38 to 0.40. That’s an astounding number and the most in La Liga by a comfortable margin. But it doesn’t offset the drop in scoring. And it’s compounded by the fact that Messi is doing less buildup work than he used to. His xGBuildup (a measure of how much he is contributing in moves that lead to the eventual xG accumulated by shots) has dropped from 1.03 last season to 0.88 this season.

This all paints the picture of a player who is beginning to focus in on the things that he’s best at while the other parts of his game start, ever so slightly, to decline. It’s always hard to define Messi’s position. But, however you look at him, either as a striker or as a midfielder/winger, the picture is the same, a player whose game hasn’t changed over the past two years, but whose production is eroding just a bit.

 

 

There are also sorts of potential reasons for that decline of course. Even the great Messi doesn’t exist in a vacuum. Grace Robertson has written about how Ernesto Valverde made stylistic changes to the squad in his second season. It’s certainly possible that it’s those changes that are slightly hampering Messi. The rest of Barcelona are also aging, and quickly. Pique, Sergio Busquets, Ivan Rakitic, and Luis Suarez are all the wrong side of 30. New young blood like Ousmane Dembele still hasn't fully settled in the side, and Philippe Coutinho has those huge theoretical, though actually probably fairly small actual, Andres Iniesta shoes to fill. It’s possible that the physical limitations holding Messi back aren’t his, but are, in fact, everybody else’s.

Even if that’s the case though, that would be a new development. One of the things that has defined Messi’s prime years, and contributed to his greatness, is his ability to but up god like numbers no matter what’s going on around him. Play him on the wing or in the center. Play him as part of a front three or a front two. Play highly possession oriented football or a swashbuckling counterattack. Prioritize defensive stability or attacking flexibility. Historically none of it matters to Messi, he’s going to do what he’s going to do and what he’s going to do is put up gigantic numbers and lead his team to titles. This year, he’s probably going to do that again, but the numbers, at least so far, are slightly less gigantic.

It’s also important to emphasize that he hasn’t even played 1000 league minutes yet this season. The blip could be just that. Maybe Messi goes off for a month, lights the league on fire, and this is just a little bit of noise that he once again successfully puts behind him. But, once players turn 30 these small downturns become more alarming than they’d otherwise be. The chance that small declines are permanent increases. Slight nagging injuries start to linger for longer. Legs that used to be fresh end up just a little bit heavier. Dips in from that a decade ago might have naturally ended quickly now stubbornly hang around. It’s too early to say that’s definitely what Messi is going through, but it’s not too early to worry that he might be.

Barcelona are still the best team in Spain. Compared to anybody else Lionel Messi is still the best player on the planet. But, compared only to himself, there’s just enough decline to start to wonder. Is this the beginning of the tail end of Messi’s career? Has he finally reached the point where instead of doing everything well, he’ll have to make choices and sacrifice some parts of his game in the service of others? Even if that’s not, in fact, what’s happening, it’s bound to happen before too long. Messi is the best, but not even the best outrun father time.

A few months into the newly formed FA Women’s Super League, it’s a good time to have a closer look to see how the biggest clubs in England have performed so far. Arsenal, who were third favourites at the start of the season, have been irresistible, stream rolling everything put in front of them. Last season’s top two, Chelsea and Manchester City, have had stuttering starts, trailing behind Joe Montemurro’s side. The race for the title and the final Champions League spot is still on so let’s take a look at the numbers to see what the rest of the season is likely to have in store. (Ed.'s note: All stats and results up to date as of November 14th)

Arsenal

Arsenal have made a blistering start to the season, winning seven out of seven. They are three points ahead of 2nd place Manchester City with a game in hand but have yet to play Nick Cushion’s side. Arsenal have managed to blow teams away this season with their impressive attacking play, scoring 34 goals in 7 matches. The most impressive result being the 5-0 win at Kingsmeadow against the double winners, Chelsea. From the above image, we can see that most of their attempts are in promising areas, mainly central inside of the penalty area. It is not surprising to see how Arsenal are able to create these opportunities as they are the most aggressive side in winning the ball back in and around the central parts of the 18 yard box which enables them to have clearer opportunities on the opponents goal, which the image below demonstrates . Most of the praise has been given to the current top scorer in the league, Vivianne Miedema. While stats for goals scored, xG per 90 and shots per 90 are impressive, her game is not just about putting the ball into the back of the net. The Holland international is integral in trying to win the ball off of the opposition but contributing to setting up chances for teammates. While Arsenal’s attack gets most of the praise, we have to also acknowledge the defensive part of their game.  Arsenal come up on top on goals conceded and number of shots conceded. When looking at the image below, we can see that Arsenal protect their penalty area extremely well, forcing their opponents to shoot from distance or put aerial balls into the box which their centre-backs are more than happy to deal with. Manchester City Manchester City, like Arsenal, are still undefeated in the league. However, they’ve managed to draw three, with probably the most disappointing one being Bristol City at home, in which they conceded two sloppy goals and the Bristol City goalkeeper, Sophie Baggaley, in inspired form to ensure they only came away with a point. It’s interesting to compare the Arsenal and Manchester City attacking numbers. Man City have generated a very similar amount of shots per game (21.33 and 21 respectively), but the league leaders have much better xG (shot quality) numbers, 3.04 per match to 1.89.   When looking at Manchester City’s shots map, it’s easy to notice why that is. When compared to Arsenal, Man City take many more shots from outside of the penalty area from various angles and more headed attempts which are harder to score from, resulting in lower shot quality.      At the other end of the field, there don’t appear to be too many issues. Ellie Roebuck has looked accomplished in between the sticks since taking the number one jersey off of Karen Bardsley while Stephanie Houghton consistently marshals her troops at the back as opponents find it difficult to create clear chances.

Chelsea

Finally, we get on to Chelsea. Emma Hayes’ side has picked up some poor results in the league despite being dominant against everyone they have faced so far in the Continental Tyres Cup and Champions League. From a defensive point of the view, they have better numbers than City and similar to Arsenal. Arsenal's opponents tally 0.60 expected goals per match while Chelsea restrict their foes to 0.53. Chelsea manage that despite allowing 8.57 shots to Arsenal's 7.17 because they restrict opponents to 0.06 xG per shot as opposed to Arsenal's 0.08. Their main concern is in attack. Chelsea have scored only 5 goals, the same number as Brighton, Bristol City and Yeovil. A few factors have contributed to that. As the reigning champions, most sides have adopted a very defensive game which Chelsea have not found easy to deal with. Having said that there are many games, like away against Bristol, where they deserved victory by a goal or two but poor finishing, brave defending and top goalkeeping have cost them dearly in the defence of their title. When comparing with Arsenal’s attacking game, there are clear differences with Chelsea’s numbers significantly lower on Clear Shots (shots generated per game when only goalkeeper was in between shot-taker and goal), xG/shot (average expected goals per shot taken by a team), 6.50 to 3.43 and Counterattacking Shots (shots generated per game within 15 seconds of a possession that originates in the originates in the opposition’s half), 1.33 to 0.43. This potentially signals that their build up play is much slower compared to Arsenal, which makes it easier for opponents to get into good defending positions to prevent Chelsea having clear sight of goal, with Ballon D’Or nominee, Fran Kirby, requiring better service. However, we must acknowledge that Chelsea are in the latter stages of the Champions League. They have played more midweek games than any of their league opponents so Emma Hayes has had to rotate her starting 11, which has potentially disrupted their game.   Conclusion Arsenal are in pole position to win their first league title in 5 years. The games vs Man City will be key and last season’s runners up will be in a good position to capitalise if they win both of those of encounters. Chelsea, on the other hand, have potentially left it too late to make a recovery for Champions League qualification via the league, but if they improve their attacking part of the game and go on a impressive run, which they are more than capable of, then second spot is still possible.   Header image courtesy of the Press Association

“The league table never lies!” and it usually doesn’t, but sometimes, there’s a bit more that can be extracted from league results, especially during the early parts of the season. If a team has faced difficult opponents, but has the same number of points as another team with a much easier schedule, then the points’ total may not reflect the whole picture. This is an attempt to produce an alternative league table which takes into account the strength of schedule faced by each team. For lack of a better name, I’ve called it the Schedule-Adjusted League Table.

The idea behind the Schedule-Adjusted League Table is to compare the teams based on their results against the same opponents. Using the Premier League as an example, take Arsenal and Tottenham. As of the 11^th of November, these are their matches played:

Arsenal Matches		Tottenham Matches
Opponent	Venue	Opponent	Venue
Man City	Home	Newcastle	Away
Chelsea	Away	Fulham	Home
West Ham	Home	Man United	Away
Cardiff	Away	Watford	Away
Newcastle	Away	Liverpool	Home
Everton	Home	Brighton	Away
Watford	Home	Huddersfield	Away
Fulham	Away	Cardiff	Home
Leicester	Home	West Ham	Away
Crystal Palace	Away	Man City	Home
Liverpool	Home	Wolves	Away
Wolves	Home	Crystal Palace	Away

Out of their 12 matches each team has played so far, there are 4 of them that are against the same opponents, at the equivalent venue (home or away). These are shown in the following table, accompanied by the number of points each of Arsenal and Tottenham have secured in each match:

		Arsenal		Tottenham
Opponent	Venue	Result	Points	Result	Points
Man City	Home	Lost	0	Lost	0
Newcastle	Away	Won	3	Won	3
Crystal Palace	Away	Drew	1	Won	3
Liverpool	Home	Drew	1	Lost	0

Of these 4 matches, Tottenham took 6 points (or 1.50 per game) whereas Arsenal took 5 points (1.25 per game). Using the same basis to compare these two teams, we might argue that Arsenal are slightly worse than Tottenham at a rate of 1.25 – 1.50 = -0.25 points per game. Let’s now turn our attention to another pair, Arsenal and Man United. Using the same procedure we find that these two teams have both faced the following 4 opponents at the equivalent venues:

		Arsenal		Man United
Opponent	Venue	Result	Points	Result	Points
Chelsea	Away	Lost	0	Drew	1
Everton	Home	Won	3	Won	3
Leicester	Home	Won	3	Won	3
Wolves	Home	Drew	1	Drew	1

Man United took 8 points (2.00 per game) from these 4 matches, whereas Arsenal had 7 points (1.75 per game). This would suggest that Arsenal are slightly worse than Man United too (also at a rate of -0.25 per game). So let’s now compare Man United with Tottenham. Same drill as before, here are their 3 common opponents:

		Man United		Tottenham
Opponent	Venue	Result	Points	Result	Points
Brighton	Away	Lost	0	Won	3
Watford	Away	Won	3	Lost	0
West Ham	Away	Lost	0	Won	3

Man United took 3 points whereas Tottenham had 6. There was also a match between the two teams which ended in a win for Tottenham, so out of these combined 4 matches Man United had 0.75 points per game whereas Tottenham had 2.25 (a difference of -1.50 points per game from Man United’s point of view). Overall, there are 190 pairs of teams in the Premier League which we need to compare. Note that the pair Arsenal-Tottenham is the same as the pair Tottenham-Arsenal.

We can compute the relative difference in points per game for each of these pairs and then solve an overdetermined system of equations (190 equations with only 20 unknowns) to find the rating for each of the 20 teams.

As there will most likely be no exact solution to the system, we estimate the solution by minimizing the sum of squared errors. More specifically, for a 20-team league, we construct W, a 190 x 20 matrix (one row for each pair of teams, one column for each team), a 20 x 1 vector x, which stores the ratings of each team (to be calculated) and a 190 x 1 vector r which has the relative difference in points gained per match. The W matrix consists of 0s apart from the column corresponding to the pair’s first team where the matrix element takes the value 1, and the column corresponding to the pair’s second team where the matrix element takes the value -1. To illustrate this, in a 4-team league, where there would be a total of 6 pairs, the W matrix would be given by:

	(Team A)	(Team B)	(Team C)	(Team D)
(Pair A-B)	1	-1	0	0
(Pair A-C)	1	0	-1	0
(Pair A-D)	1	0	0	-1
(Pair B-C)	0	1	-1	0
(Pair B-D)	0	1	0	-1
(Pair C-D)	0	0	1	-1

Essentially, the system that we need to solve is defined by Wx = r. As there will generally be no solution to the system, we can approximate it by minimizing the sum of squared errors, which leads to the ratings solution vector given by:

x = (W^TW)^-1W^Tr

where the inverse is a generalized inverse. Once the ratings solution vector x is calculated, a suitable scaling is applied – huge thanks to Christopher D. Long (@octonion) for the help provided in this part – and the schedule-adjusted points for team i is given by:

AdjPts_{i =} m_i * (x_i + TotPts/2n)

The results from this analysis in tabular and graphical form are presented below (after 12 Premier League matches):

	Actual Points	Adjusted Points
Man City	32	33.67
Liverpool	30	30.94
Tottenham	27	29.57
Arsenal	24	26.61
Chelsea	28	26.44
Man United	20	22.4
Everton	19	19.5
Watford	20	18.46
Wolves	16	18.04
Bournemouth	20	18.03
Leicester	17	17.22
Brighton	14	14.54
West Ham	12	13.86
Crystal Palace	8	9.32
Newcastle	9	7.18
Huddersfield	7	7.11
Burnley	9	6.71
Southampton	8	5.87
Cardiff	8	5.45
Fulham	5	2.08

This methodology would suggest that teams like Arsenal, Tottenham, Man United and Wolves have registered points’ totals which underestimate the true value of those wins and draws. On the other hand, the value of the points gained by teams like Fulham, Cardiff, Burnley and Southampton is in fact lower, once the strength of teams they faced is taken into consideration. We can also view how the schedule-adjusted points compare against the actual points on a gameweek-by-gameweek basis in the following animation:

In a sense, teams above the 45-degree orange line are short-changed as the true worth of their points gained is more than the league table would suggest, whereas teams below it may be viewed as having a deceptively high number of league points given the true worth of those points. Alternatively, one could view the teams’ ranking on the horizontal axis as the actual league table and the teams’ ranking on the vertical axis as the schedule-adjusted league table.

As the season progresses, the points will approach the diagonal, before lying exactly on it in their final ranking, whichever that may be. Some additional notes on the methodology:

• It does not depend on any prior beliefs regarding the strength of each team. Therefore, a win against a presumed weak team which however has been getting results will contribute more compared to a win against a traditionally strong team which hasn’t been getting any league points.
• A team’s adjusted points might change even if that team does not play a match, conditional on other results. This means that the adjusted points gained from a win against a particular team might increase or decrease, depending on whether that particular team has been winning or losing subsequent matches. This provides a continuous update on the true strength of a team and therefore adjusts the value of points already gained by other teams against it.
• Any concept of “form” is not considered. As a result, the schedule-adjusted table would still be the same, even if we re-arranged the playing order of the matches up to that point.
• By construction, all teams are considered as equal at the start of the season hence the schedule-adjusted league table will appear compressed compared to the actual league table in the early stages of the season. This is corrected once the number of pairwise comparisons between teams increases, as that will provide additional information on the relative strength of each team.
• The schedule-adjusted table converges to the actual league table once all matches have taken place.

It should be stressed that this methodology can be extended and applied to other similar situations irrespective of the metric used. For example, given its recent popularity, one could use it to assess the true worth of xG figures registered by teams, once the difficulty of the opposition has been taken into account.

So there you have it!

This piece provides the idea and the methodology to produce schedule-adjusted rankings. Next time someone points at the league table to support an argument, get your calculator out and check if that table is indeed telling the truth.

Happy schedule-adjusting!

It’s Thanksgiving in America. That means turkey, fights with relatives about politics, and gimmicky content. With that in mind, here’s an entirely incomplete list of things I’m thankful for across soccer this year.   Attackers getting healthy Callum Wilson Wilson has been the catalyst for Bournemouth’s surprisingly strong start. The team sits in sixth with 20 points, and has an expected goal difference of 0.33 per game, also sixth best in the league. Wilson is a huge part of that. He’s one of only seven players (with over 500 minutes played) in the Premier League who are averaging over 0.30 expected goals per match and 0.20 expected assists. The rest of the list is Sergio Aguero, Riyad Mahrez, Raheem Sterling, Ross Barkley, Mohammed Salah, and Raul Jimenez. In other words, he good. He’s not a particularly high volume player, but when he does touch the ball he turns the opportunity into an extremely high quality shot for himself. Of the league's higher volume shooters (those averaging over two shots per game in 500 minutes played or more), he’s one of only two players to average over 0.20 expected goals per shot. He’s at 0.20, Pierre-Emerick Aubameyang just noses him out at 0.21. Wilson is coming off of two separate cruciate ligament ruptures, one in 2015-16 and one in 2016-17. Last year he was finally healthy and managed to start 23 Premier League games and appear as a sub in five others. And while he only managed eight goals and two assists it’s certainly reasonable to hope that last year was about getting back to former form and this year marks a return to the production he logged before his knee blew up a couple of times. When he injured himself in his first Premier League season, Wilson was coming off back to back 20 goals seasons, first with Coventry City in League One and then with Bournemouth as they won the Championship. Hopefully, especially for Bournemouth, this season just marks the slightly delayed arrival of Wilson at the top level that those two years portended.   Marco Reus He’s always been great. He’s always been injured. Now that he’s back and firing on all cylinders he’s the scoring machine from the wing (and sometimes the front) that Dortmund’s title challenge is built on. He's 29, he probably doesn't have too many hurrahs left, so if this is his last one, hopefully it's a good one.     Helter skelter La Liga There are only seven teams in Spain who concede more expected goals per game than Barcelona. They are hanging on to their first place position not because of anything they’re doing on the defensive side of the ball, but because Lionel Messi is leading a line that can (and has to) outscore anybody and everybody. The shots that defense gives up are pretty pretty ugly looking. It’s not just Barca either. Second place Sevilla has leaned into attack at the expense of defense. Alaves in fourth place is an even worse defensive team than Barcelona (Alaves also has the third worst expected goal difference in the entire league at -0.46, so just exactly how they’re in fourth place remains a mystery). Even the vaunted Atletico Madrid fortress is conceding over an expected goal per game. Only Getafe, Eibar, and Valencia have managed to hold opponents below the one expected goal level. La Liga is weird this year. There’s upheaval at Real Madrid, weakness at Barcelona and some traditional second tier competitors like Valencia and Villareal have played well but struggled to get results. Sometimes weirdness leads to defense and boredom, but thankfully in La Liga this season it’s leading to good old fashioned end to end stuff.   The Burnley keeper experiment Last season Burnley were magic. They finished seventh. Part of what happened was Nick Pope did this. He was great. Last season Joe Hart played for West Ham. He was terrible. So naturally when Pope got hurt, Burnley went out and bought Joe Hart and handed him the keys, as you do. So far, he’s been…entirely average. What does it all mean? I have no idea! But this experiment should help us figure it out.   Young stars breaking out Andre Silva Everybody thought Andre Silva had potential. After announcing his presence at 20 years old for Porto with 16 goals in 32 games (albeit with five from the penalty spot) he promptly moved to AC Milan where, for whatever reason, he fell on his face. He only started seven games and scored two goals. A year later, he’s still only 23 years old, and he’s a fixture of Sevilla’s high-flying attack. He already has seven goals and his 0.52 expected goals per 90 is fourth in La Liga among players with more than 500 minutes. He’s not the highest volume shooter, 3.03 is tenth in the league, but his 0.17 expected goals per shot is fourth in La Liga among high volume (more than two shots per game) shooters. He camps out in the middle of the penalty area and puts it on net, and that’s exactly what Sevilla needs.   Luka Jovic He’s 20. He’s scored nine goals in less than 600 minutes. Apparently all he does is shoot. It sure seems to be working.   Jaden Sancho Whooooo boy.   Happy Thanksgiving, enjoy your Turkey.

Sevilla are riding high in La Liga on the back of an excellent start to the season. Powered by the second best attack in the division, they have good underlying numbers to underpin what will likely be a genuine challenge for their first top-three finish in a decade. It is all far removed from the madness of last season. That summer, coach Jorge Sampaoli and long-serving transfer guru Monchi departed. Sampaoli’s replacement Eduardo Berizzo was gone before December, his replacement Vincenzo Montella lasted until the end of April, and Joaquín Caparros then stepped in for the run in of three wins and a draw that rescued a place in the Europa League qualifiers. Oh, and they also reached Copa del Rey final and the last eight of the Champions League. In the midst of all that, a finishing position of seventh, with the league’s seventh-best xG difference, was a pretty decent return. After his temporary reign on the bench, Caparros moved into a director of football role, and turned to Pablo Machín, fresh off an accomplished debut top-flight season with Girona, to lead Sevilla forward. Machín’s team weren’t quite as impressive from a numbers standpoint as fellow newly promoted side Getafe - who again look very solid this season - but their intense and direct approach yielded results and was a good fit with Sevilla’s traditional style of play, one that Caparros himself helped forge as head coach between 2000 and 2005. Girona pressed high, regularly won the ball in the middle of the pitch and then transitioned swiftly to attack, with plenty of diagonal passes out to their rampaging wing-backs and crosses into the box. They were also very effective on set-pieces, with a league-high 38.64% of their non-penalty goals coming from such situations. They generated over 50% of their non-penalty xG from a combination of set-pieces and open-play crosses. It was all relatively simple yet well-drilled - the culmination of Machín’s four years at the helm. With a few tweaks that approach has so far scaled well to a bigger club. At Sevilla, the 3-4-2-1 formation Machín utilised at Girona has given away to a 3-5-2 (perhaps a 3-1-4-2 or 3-5-1-1 depending on how you wish to split the bands), clearly visible in this pass map from last month’s home victory over Celta Vigo. At Girona, the wing-backs were usually the first attacking out ball, maintaining high positions to stretch play horizontally and thus create space infield for the two attacking midfielders, who operated almost as classic inside forwards. At Sevilla, however, his use of two forwards, with André Silva generally playing just off of Wissam Ben Yedder, occupies the opposing central defenders, stretching play vertically and generating space for the two advanced central midfielders to attack. The wing-backs normally join in a bit later. The identity of Sevilla’s regular midfield starters makes it clear that the focus of the side is on attack. A trio of Pablo Sarabia, Éver Banega (getting regular minutes as a lone defensive midfielder for the first time since his early days at Boca Juniors weaving wonderful passing patterns with Juan Riquelme) and Franco Vázquez is certainly not one constructed to provide defensive resistance, as their numbers for the season to date attest. Sevilla’s xG conceded has marginally improved from last season, to a league-11th-best 1.18 per match, but their attacking output has shot up from 1.33 xG per match last season to 1.80 this time around, driven by increases in both shot volume (15.08, up from 13.66) and quality (0.12 per shot, up from 0.10). A side who were passive and one-paced for large stretches of last season are now direct and incisive. After a barren year at Milan following his big-money move from Porto, Silva has quickly made an impact at the Sánchez Pizjuán. Not only is he scoring goals (seven, off 5.72 xG), but he has impressed the coaching staff with his ability to hold up and link play. His strike partner Ben Yedder is also very much enjoying himself. The ever-productive Sarabia is another thriving in this setup. As the right of the two advanced central midfielders, he is able to turn infield onto his favoured left foot and enjoy a propitious panorama, with Jesus Navas advancing wide to his right, two forwards ahead of him and other options emerging to his left. He already has four goals and four assists, and ranks towards the top of the team in terms of deep progressions, key passes and on the other side of the ball, pressures and pressure regains. Sevilla’s defence was a clear weakness last season - they conceded three or more goals on seven occasions - and Machín has simply chosen to embrace the talent imbalance in his squad. His team generate plenty of opportunities, but it is also relatively easy for opposing teams to advance through the center of the pitch and get off shots against them. Their matches have averaged a pretty evenly distributed 30 shots. Machín is betting that by throwing numbers forward his side will create better quality chances than they give up. It is a risky strategy but one that’s working well so far. Sevilla are currently just a point shy of leaders Barcelona following a hard-fought, come-from-behind victory at home to fast-starting Espanyol prior to the international break. There is little in their underlying numbers to suggest that form is unsustainable; on xG difference, they are La Liga’s second-best team. But there are some question marks. Have they benefited from the fitness boost of getting into competitive action earlier than others due to their participation in the Europa League qualifying rounds? Will that catch up with them later in the campaign, especially given that six outfield players have started at least 11 of their 12 league matches to date? Will they fall off like Machín’s Girona side did last season? Girona ran a positive xG difference of 0.20 per match through the first half of the campaign but that dropped into negative territory at -0.26 per match thereafter, with regression at both ends of the pitch. Whether that was due to tiredness (he again ran his favoured starters hard) or opponents figuring out their approach, it is something Machín will need to guard against this time around. Conversely, he still hasn’t been in charge for all that long. Sevilla aren’t yet the organised pressing unit his Girona side were, while their set-piece productivity is also likely to improve with further practice of routines and strategies - something that various members of the squad have highlighted as a focus. Barcelona have so far been a step below their level of last season, opening up a window of hope for Sevilla in terms of a title bid. The top three is probably a more realistic aim given Real Madrid’s relative struggles and the fact that Atlético Madrid, currently level on points with Sevilla, are having real problems in attack. At the very least, Valencia’s poor start - in terms of results rather than underlying numbers - has opened the door to the top four and Champions League qualification, which in of itself would represent a good return from Machín’s debut campaign. Header image courtesy of the Press Association