StatsBomb: Round Table Q and A

excalibur2 So we reach the business end of the season. By way of contrast to the many models we see, we thought it might be interesting to gauge opinion from our writers and a small group of friends of the site. It’s important to remember that although every one of the contributors has done extensive writing or work with a statistical slant, they are all first and foremost football fans. And what do football fans have? Opinions. Cobbling together a few questions was straightforward enough and a mix of reflections and predictions should be enough to stir a mixture of agreement and horror. What do we learn? Stats guys like:

  • Romelu Lukaku
  • Barcelona and Dortmund for the Champions and Europa League (Innovative!)
  • Bournemouth to improve next year
  • Leicester to decline- markedly

And plenty more. Enjoy! Contributors Paul Riley (PR), Mohamed Mohamed (MM), Dustin Ward (DW), Will Gurpinar-Morgan (WGM), Clarke (C), Thom Lawrence (TL) Peter Owen (PO), James Yorke (JY) Full profiles are at the end of the article.


1. Who wins the Premier League and why? WGM: My expected goal numbers favour Arsenal overall, so I’ll go with them. Really though, there is still a long way to go and we’ll likely see plenty of twists and turns in such a tight race along the way. PR: I don’t know! The only team I’d rule out at this stage is Leicester as I don’t think they have the quality or experience to last out. What a run they’ve had, though, it’s been brilliant. PO: Manchester City.  With Kompany, Toure, Silva, and Aguero all fully fit City have the best team in the league. City also have a much nicer run in than Spurs and Arsenal. If either one of Kante/Mahrez/Vardy gets injured then I think it puts Leicester at a significant disadvantage to the rest of the top 4. C: It is close but I have Spurs wanting it more than the other 3 contenders. Pochettino has done a great job of wrangling the confidence away from other teams while at the same time instilling a mental toughness that really shines through in set piece situations. DW: Arsenal. Significantly better than Leicester and Tottenham and 4 point edge on City. TL: Tottenham would do nicely. No drama, the most reliable machine in the top 4, and I struggle to find a good reason they _shouldn’t_. MM: Arsenal will win the league. The full team coming back healthy for the stretch run seems like a good omen for their shot numbers to stop diverging from their expected goal numbers. Getting knocked out of the Champions League by Barcelona will mean that it’s one less competition to deal with. A fully functional Arsenal, which can be as rare as a meteor shower, is the best team in the league this season, and I’m betting that we’ll see that for the majority of the final dozen games. JY: Tottenham are the best team in the league, Leicester will drop off, City are asleep and Arsenal still aren’t clicking. So there.


  2. Player of the year so far and why? PR: Riyadh Mahrez. The way he’s maintained his form throughout Leicester’s season is pretty amazing considering the hurdles he has to jump that other contenders don’t: firstly, Leicester’s switch in tack from toe to toeing to rope-a-doping, secondly, the direct style this takes in, and thirdly, the level of player around him. Young Player? Harry Kane. Leading the line, mixing it up like he does and still getting that no of shots off at his age is really quite something. I love the way he strikes a football. PO: N’golo Kante. Has to go to a Leicester player really doesn’t it and Romelu Lukaku (yup, even after multiple generations in the premier league he’s still young enough to win it) TL: Lukaku for young player, lest you need reminding that he’s 22. In a way, he’s rivalling Lineker’s 30 league/40 total goal season, scaled back to a far worse Everton team than the mid-80s incarnation. C: Ozil and Lukaku. Pretty obvious reasons, I think. Ozil is trying this year and Lukaku is gaming expected goals. DW: Mesut Ozil and Dele Alli. Probably win it on even playing field but the fact he is only 19 pushes him further ahead. MM: Paulo Dybala. Had Nabil Fekir not torn his ACL he could’ve been a candidate and Lorenzo Insigne until about late December was having a banner year for a resurgent Napoli, but Dybala is the best player on Juventus and over the past 3-4 months has been the most dynamic striker in Serie A. He’s much better than I thought he would be right now and even at the price they paid to Palermo, Juventus got a great bargain JY: Ozil narrowly ahead of Mahrez, think his crazy high assist total is enough to swing it for me, and I like the idea that he’s owed some props from the UK press who typically haven’t “got” him. Young player? Whole shortlist could be Tottenham players, Lukaku is still young enough too. Been a good year.


  3. Manager of the year so far? PR: Pochettino. Spurs were a bit of a shambles last season. Now I can’t really see many weaknesses at all…and the numbers back that up. PO: Again feel obliged to go with Leicester so it’s gonna have to be Ranieri. C: I refuse to give this award to the manager of the team with a high PDO. (Sorry, Claudio, Mauricio, Arsene and Slaven). In a wildly unpopular move I am voting for Manuel Pellegrini with Man City still top two for a ton of metrics and all that with an injury riddled line-up. He hasn’t got the positive skew like the others and is still in striking distance. Plus, he seems like a nice person. WGM: Ranieri. While Leicester appear to be have hitched a ride on a tidal wave to propel them to first place, it’s important not to go too far and put all of their improvement down to the footballing gods. DW: Ralph Hasenhüttl at Ingolstadt. Has taken a team with relegation-level talent and relegation-level money solidly into the mid-table by turning them into a ferocious defensive machine. Ugly with the ball but effective overall. MM: Pochettino. I’ve liked him going back to his Southampton days and what he’s done with a Spurs side that’s soon to be transitioning into a new stadium with a ton of young players has been remarkable in such a short time. Hopefully for Tottenham, he won’t be swayed elsewhere this summer. TL: Ranieri. Not since David O’Leary has a manager taken such a scrappy team of misfits and racists, skipped training a couple of days a week, and broken through the glass ceiling with such ease. JY: Even beyond my obvious bias, I think Pochettino has improved Tottenham more than Ranieri has Leicester and a scratch at the numbers pretty much supports that, though I accept it’s arguable.


  4. Which player from a bottom half team should bigger clubs look to sign this summer? PR: Romelu Lukaku. He’s wasting his time at Everton. He could be a huge star. TL: The way things are going, probably Lukaku. PO: Diego Costa WGM: I was going to suggest a Chelsea player but figured that was too easy/troll-like. So I’ll go with Lukaku. DW: Alexandru Maxim from Stuttgart. The 25-year old isn’t getting a lot of starts but consistently puts up incredible passing numbers from midfield. If you can carry his lack of defensive work, he’s great going forward. They might not be willing to get rid of him as cheaply as his playing time suggests with Didavi a free agent but it’s worth a call. MM: Lukaku. He’s one of the 10 best strikers in the world already and he’s not even reached peak age. For as much hype as Michy Batshuayi has garnered, Lukaku is just a few months younger than him with four years of PL goal scoring to his resume. We could live in a world where he is the best striker in the world. JY: Hmm… depends where Everton end up, Barkley and Lukaku are at the right point on the curve for a big club. As you can see from the other answers, this isn’t a simple question, most of the cream is already at the top.


  5. Who should be most disappointed by this season? Man Utd/Chelsea/Liverpool/? PR: They should all be thoroughly ashamed. They all lack any kind of proper leadership off the field. WGM: It’s hard to look beyond Chelsea given the historic nature of their fall. PO: Going with the easy answer and saying Chelsea C: Dear, Liverpool supporters: process should always outweigh results and the process looks pretty good. An 18 month PDO slump can really wear you out but understand better days are coming. DW: Manchester United. They have the most money and the most stability and have arguably performed the worst. Chelsea and Liverpool have certainly been unlucky to be so low in the table, to what degree we can debate. Manchester United have been pretty lucky to be as high in the table they are. MM: Chelsea just because of the magnitude of their drop this season. Liverpool were always flawed and LVG constricts shot numbers like he has a personal bet with someone. Chelsea were the only team that had genuine aspirations and they probably won’t have European football next season. JY: Man Utd. They aren’t unlucky, and it hasn’t even been entertaining for the neutral. That’s criminal to me. TL: Chelsea, with a truly historic collapse brought about by everything going wrong both on and off the pitch.


CcFAWb6W0AAPlIv.jpg large 6. It is the Champions League Final, a quirk of fate has meant that your life depends on beating Barcelona by selecting one Premier League team to beat them. Who do you pick? PR: Tottenham. Pochettino’s the only manager I’d back to come up with a gameplan to give them a game with the players he’s got. PO: A fully fit City with Pep at the helm. C: A full strength Man City. TL: Arsenal. I believe their fans are the most psychologically conditioned for defeat by Barcelona so from a utilitarian standpoint it would probably cause the least anguish, despite my inevitable doom. DW: Manchester City. It’s either Arsenal or City as those two have too large a quality gap with the rest for matchup concerns to factor in. I think City healthy is better and don’t see any sort of tactical or matchup reason that Arsenal would do much better so I choose the best team. MM: West Brom. I’m going to die anyway if I pick a Premier League team to beat the best team in 20-25 years so I might as well live and die by Pulis ball. JY: I always think a fully fit Arsenal can give anyone a game. [written pre-Champions League game, and that didn’t pan out]. Obviously, THIS SPURS TEAM would be comfortable. WGM: I really like Spurs’ defense this year with the combination of pressing higher up the pitch and in midfield combined with the protection they offer to their central defense. That ability to be proactive in defense could serve them well in advance of Dele Alli’s winning panenka penalty in the shoot-out.


  7. Disappointment of the year (player/transfer)? PR: None, really. It’s been a good year overall hasn’t it? And all transfers come with risk attached and I rarely agree with any that are made anyway! Christ knows what’s in the thinking of managers/clubs and their recruitment policies. A lot look like random acts of kindness to the seller! PO: Jordon Amavi – not disappointment with his performance but with the fact we didn’t get to see him play more, and also that he’s probably not going to be playing in the Premier League next season, poor fella. C: Many Chelsea players, Memphis Depay and Steve McClaren TL: Aston Villa’s entire Summer window, not because I think they did _egregiously_ bad business, but because it’ll go down as cute analytics signings sinking one of England’s most venerable clubs. DW: Josip Drmic. It was easy to see coming that replacing Max Kruse with Drmic was going to be a huge downgrade for Gladbach. Drmic is bad with the ball and floats on the periphery, 10 million euros for him was too much. He played 400 really bad minutes and is already off to Hamburg. MM: Jackson Martinez to Atletico. There were warning signs that this wouldn’t work well but it shouldn’t have happened this quickly even with Martinez’s age. Luckily for Atletico, China came calling and they somehow ended up making a profit. JY:  I feel sorry for Depay. Soon as he picked Utd I thought it was a bad idea. He’s young and can get it back but living with Louis is miserable. [edit: has game of his life night before this article released]


  8. Are you happy with your team’s season? What is the first problem they need to solve in the summer? PR: (Everton):  Nothing gets sorted at Everton without new owners and some actual plan off the field beyond existing. PO (Liverpool): Results, nope. Although getting Klopp has made everything worthwhile. C: (Middlesbrough):  Yes, I am pleased. I don’t want to jinx it but our GK has been confident. DW (Liverpool): Yes. Hiring Klopp makes me happy, I enjoy his personality, his antics and like that he seems to be able to connect emotionally with players in a way that Rodgers didn’t seem to do. This summer Liverpool need to get a lot more athletic. There are big physical weaknesses everywhere but the largest might be speed in the midfield and defense. Seeing Lucas and Kolo Toure try to cover large open spaces is not pleasant. MM (Marseille): As the only one here to support a broke French club, I’m not happy at all with this season. Michy and Lassana Diarra’s revival have been the two saving graces and Michy will be gone in the summer. I don’t want to go on and on about this but Marseille should stop hanging by the coattails of big clubs/PL sides for loans and start taking advantage of France’s vast amount of produced talent from smaller clubs/lower leagues, which they’ve done previously but have forgotten in recent windows. JY (Tottenham): Yep, I’m still at the stage where I will take top 3/4 as it stands, anything else is a bonus but next year the first problem is going to be quelling expectation that we will bulldoze the Champions League. It’s been genuinely exciting and surprising to see such a dramatic improvement over such a short space of time though, I figured next season was when it could come together if it all went right, and we’re there already. Great stuff. TL (Everton): Nope. Our defence continues on the standard Martinez trajectory, and we have an incredible, free-wheeling attack that has failed this season to crack even the smallest nuts when it actually counted. I’d be intrigued to give Martinez one more season to prove he can build a system around Stones that actually functions, but I assume the transfer market will blessedly take the decision out of his hands. I’d personally accept we’re not close to the top 4 and maintain cool heads while working out long term replacements for Howard, Jagielka, Baines and Barry, while finding some depth at right-back and in our attacking positions. WGM (Liverpool): It’s hard to evaluate the season. In isolation, it’s clearly been a disappointment in terms of results but the hope is that we’ll look back on this as the start of a successful era under Jürgen Klopp. There have been some encouraging signals in some of the underlying numbers but the team was clearly hampered by the injury crisis and fixture schedule after a promising start to Klopp’s reign. A good run until the end of the season would be nice though!  As far as the summer goes, the key for me is that everyone at the club works together and pulls in the same direction in relation to transfers. After Rodgers’ sacking, the hatchet jobs in the media on the transfer committee, and Michael Edwards in particular, made it abundantly clear just how dysfunctional the club was in terms of the recruitment process. Mistakes will still be made no doubt but having something approaching a professional process would be a vast improvement.


  9. Where will Leicester finish next season? PR: Mid-table obscurity. TL: Bold prediction: top half. They never needed depth this season, but will now have the cash to plug that hole without anybody noticing. The crown jewels seem to be staying and I’m sure their analytics peeps have an intriguing Summer planned out, but I think with a more packed schedule and a normal amount of luck, they are not going to repeat their dominance in a cash-rich, Pep-powered league. PO: lol C: Enigmatic Leicester. So many factors at play. Will key players stay for the Champions League? Can they attract better players? Will the extra games affect the league? Plus, will their style of play, high conversion rates and a lack of injuries catch up with them? I will go with 7th but I wouldn’t be that surprised if they end up 14th. DW: A fair over under for me would be 9.5th place. MM: Top 8. I think the majority of this squad will come back and they’ll realize that a lot of variance both statistically and fixture wise went their way. From the looks of it, Leicester are a forward thinking side so they’ll try and add to the squad with the extra money coming their way. There’s like a 3% chance that Leicester can somehow become a semi permanent top 4 club with smart squad building and the influx of money but I’m going to go with the 97% and a top 8 finish seems more likely. JY: 8th. They have no squad. Yeah, they’ve been good but probably need to hang onto Mahrez (unlikely) and buy about 6 solid players just to stand still. And standing still, really, is edge of Europa pace.


  10. Which club is best positioned to “do a Leicester” next year and improve into the top 7? PR: Chelsea!? PO: With a couple good centre mids Everton could do something very similar to what Spurs have done this season,not quite what you’re looking for but close enough. DW: Bournemouth. I really like Eddie Howe and they dominate territory this year more than teams like Manchester United and Everton so the bare bones are there for something really good. Maybe controlling territory somehow caps “low-talent” teams ceiling but I generally doubt that and think they are building something impressive. C: I think Bournemouth could surprise. Hindered by injuries and a crazy low save% but still top 10 in a variety of shot metrics. MM: It’s actually hard to answer this question. Southampton doesn’t really count because they’ve been a top 7-8 side the last two seasons and flirted with a top 4 finish in both seasons,  Stoke have nice players and a commendable project but I’m not a fan of Mark Hughes and I’m not trusting Pardew period. So I guess…  let’s go with Everton. If they can improve the shambles behind their front four, find some ball playing midfielders, find some young CBs that can play and keep Lukaku/Barkley/Deulofeu together, maybe they can make another miracle run like they did in 2013-14. JY: This depends on how we define “doing a Leicester”as I think so much has to go right to even get a sniff at the edges of top 7 and with the fallen giants surely regrouping, it’ll get no easier to break in. Throw in your Stokes, West Hams, Southamptons and 10th suddenly looks a good finish. Maybe Swansea hire someone a bit fun and get the magic skew again like last year. TL: Everyone’s going to have so much money it seems unlikely we’ll experience such a surprise again – even the teams in the Championship promotion places are splashing the cash. Obviously if Leicester didn’t exist we’d be raving about Watford and perhaps Ighalo can power them to another strong finish. The only fairy tale that remains available is really Bournemouth, so it’d be wonderful to see them avoid some of the bad luck they experienced early this season.


  11. Do you sell Daniel Sturridge in the summer? Do you buy Daniel Sturridge in the summer? PR: Liverpool shouldn’t have to sell. They’ve got enough money to buy as well as, not replace. Conversely, why would I buy someone that injury prone when I’ve got enough TV money now to go shopping anywhere else in the world? Unless the price is right of course… WGM: I’m totally biased but when fit, there are very few strikers around who are better than him. The list of strikers that are better than him and who Liverpool could conceivably buy is even shorter (I’d argue it’s pretty much zero). ‘When fit’ is the obvious caveat but if Liverpool can put together an appropriate fitness programme/treatment, then they’ll have a player who is much better than what they currently have available. Investing in a younger proto-Sturridge with potential to develop should be doable in the summer with funds available plus current fourth choice striker, Christian Benteke, could be sold for a reasonable sum of money (plus Balotelli leaving would free up wages). PO: Honestly couldn’t think of anything more stupid. Liverpool would be selling their best player, probably at a lot less than true value. His wages are nothing at a macro level, and when he does play it improves the team more significantly than any other player in the squad. If Liverpool want to be a top 4 team their best bet would be keeping hold of Sturridge. TL: The ultimate glass cannon, he has pretty irreplaceable numbers when he plays, so depending on Liverpool’s bank balance I’d probably just keep him around as a luxury. I think it’d be a difficult sell for any club’s fans if you were to drop serious money on him, especially for teams that are in the middle of rebuilding anyway. Then again, if you’re someone like Man City, near their peak, and looking to plug gaps in elite players’ seasons like Aguero, you might hope that Sturridge’s injury schedule gives you one whole elite striker in the aggregate. Big gamble, but his numbers are real. C: The answer to both questions is, ‘Damned if you do, damned if you dont.’ Doesnt play enough to warrant his wages but when he does play he is worth every £. DW: No and no. If he’s healthy, he’s Liverpool’s best player. There don’t seem to be any serious structural problems so I don’t think they should sell him. If they do sell him, as a buying team I’d be suspicious there is either a structural problem or Liverpool believes he will just never trust his body again. If the team that knows him best puts him up for sale, that’s a data point against him. MM: Yes and maybe. I think Daniel Sturridge when healthy is easily the most dynamic player Liverpool have. Yes Suarez helped elevate Sturridge to a different level but even without him, he’s been productive. I’m just super skeptical that he can stay fit and play 1800-2000 league minutes. JY: No, cos then you have zero top class strikers instead of half a broken one.


  12. Who wins the Champions League/Europa League? PR: Barcelona/Sevilla PO: Gonna have to say Barca and Dortmund C: Clubs from Spain DW: Barcelona or Bayern Munich are the two best teams by far but if I had to bet against public opinion on my twitter timeline, Real Madrid. Generally underrated by analytics community. In the Europa League Dortmund are by some distance the best team. TL: It’d be nice if Bayern Munich stayed fit enough to make it a fair fight, but Barcelona are presumably the safe bet. I was looking forward to Valencia vs Midtjylland in the Narrative Cup Final, the Danes winning on penalties after discovering a previously unknown loophole in the laws of the game allowing them shoulder barge the keeper into the net, but it’s not to be… WGM: Barcelona and Borussia Dortmund. MM: Barcelona/Dortmund. If Dortmund were playing in the CL, they would be regarded as one of the 4-6 best teams in Europe so I think they’re prohibitive favorites to win the EL as long as they try. Barcelona are one of the best teams to have ever graced the Champions League so unless multiple injuries happen to key players, I find it hard to believe that they won’t repeat. JY: Boring, boring Barcelona, who are clearly anything but and er… Sevilla? Wait, what am I saying? Tottenham Hotspur is once more the answer.


  13. Anything you’d like to make comment on that hasn’t been covered by these questions? DW: I think that it’s very likely that within 7-8 years there will be 4 or 5 Chinese teams better than most teams in the French League. PR: Good lord, no. JY: With the news this week that Roy Hodgson hates stats, we must redouble efforts to infiltrate the mainstream.   Thanks for reading! Contributors Paul Riley (@footballfactman)

Long regarded as a foremost analytical evaluator of goalkeepers, Paul has shared his expected goal dashboard and chance creation maps this season with the site. In a rumoured final foray he has recently started work on a series of analytical analyses on the forthcoming England 2016 squad.

Mohamed Mohamed (@MoeSquare) StatsBomb’s resident Ligue Un expert, Moe has recently branched out to Spain with a look at Rayo Vallecano and the lower reaches of the Premier League with Stoke and Aston Villa, amongst other work. He has written articles across a variety of tactical and analytical websites and recently started a podcast based on his first love, Ligue Un. Dustin Ward (@SaturdayOnCouch) StatsBomb’s resident Bundesliga expert, Dustin has produced series of popular articles on the league and has become renowned for his deep dives into the numbers, often revealing sharp insights. He has also recently looked at the defensive structure of Atletico Madrid and soon a long awaited shift to the Premier League will lead to an analysis of his own team, Liverpool. Will Gurpinar-Morgan  (@WillTGM) A former Optapro forum presenter with a strong technical and statistical base, Will has been posting analytical work since 2012 on his blog. Most recently expanding on his non-shot expected goals model to look at a variant of plus/minus, he has also recently evaluated aspects of Arsenal’s season and looked at differences in metrics via shot expectation. Clarke  (@footyinthecloud) Running the essential “Football In The Clouds” website, Clarke provides a unique take with his 11 v 11 shot numbers as well as advanced on pitch individual player stats. Peter Owen (@_PeteOwen) In looking at football through a statistical lens, Pete is a valued contributor here at StatsBomb. He has recently evaluated aspects of luck versus skill, looked at officiating and competitive balance. Thom Lawrence (@deepxg) In a whirlwind entrance to analytics, Thom under his Deep xG moniker has quickly won many plaudits for his ideas and visuals. Most recently building his PATCH defensive metric, his website is full of innovation and those ubiquitous polygons. James Yorke (@jair1970) Editor of StatsBomb, as required, and writer of a weekly Premier League Round Up column. Also recently started contributing for @ESPNInsider    

Luck vs Skill And Attack vs Defence

In a sport where one goal being scored is the most common result, luck is undeniably going to be a huge factor in influencing results. There are a few different ways of looking into how much look is involved in football. PDO is widely used in the football analytics community to determine whether an individual team is over or under performing by looking at how frequently shots on target are being scored/saved. I’m going to look into a method that was suggested by Tom Tango that has been adapted to work out how much variance you would expect in a completely random league, and how much variance actually occurs. I’m going to follow on from analysis by Julian Ryan who wrote this piece on Harvard Sports Analysis.

This technique has been used before by James Grayson and Martin Eastwood in order to find the amount of luck involved in win% for Premier League and La Liga seasons. Both Martin and James found that the % of skill in points for a given season is about 60-65%. Instead of looking into the variance of win%, I’m going to venture into the variance in scoring and conceding goals and see if there’s much or a difference between win% and goalscoring in the Premier League.

Calculating skill%

slightly maths-y/technical bit coming up so if you want to skip to the conclusions be my guest.

To figure out how much the variation in teams’ results is due to skill we can do the following calculation:

Observed variance = Variance due to skill + Variance due to luck

Where variance due to luck is the amount of variance we would expect if all teams were equal and games were played out at random.

We know due to various studies that goals follow a poisson distribution and we can therefore model the number of goals scored this way.

Let’s see how we can model our luck distribution.

The variance of a poisson distribution is equal to the mean of the distribution. The mean/average number of goals for an EPL team in a season is 50.2. Let’s assume that for each game our “luck team” will expect to score (50.2/38) = 1.32 goals.

if xG1 is the expected number of goals for a team in game 1, and xG2 expected number of goals in game 2 etc. and we assume the expected number of goals in each game is independent, the following formula holds:

Var(xG1 + xG2 + xG3 + …) = Var(xG1) + Var(xG2) + Var(xG3) + …

Since the expected number of goals in our “luck distribution” is the same for each game, and we are assuming each game is independent, we can simply sum up the expected number of goals for each game to find our total variance. So:

(50.2 / 38)  *  38  =  50.2

which would be the variance of our luck distribution. For Goal Difference we can use the variance formula again to see that

Var(GF + GA) = Var(GF) + Var(GA)

Therefore to get our GD variance for our luck distribution we can just double the variance we get for GF or GA.

We can calculate total observed variance from the premier league table using using the usual formula for variance, depending on whether we’re using a sample or the population. We’re going to be calculating skill% where: Skill% = 1 – (Variance of our luck distribution/Total observed Variance)

Results

Below is the distribution of skill% for goals for, goals against and goal difference. Data is from the past 15 Premier League seasons.

 

luckdist3

 

From season to season the amount of variation explained due to skill can vary significantly. On average though, a teams goal difference is the measure that contains the most “skill” and goals conceded has the largest amount of “luck”.

If you look at the minimum values of skill% you can see that spread of GF can be pretty random over a season and the same with GA. However, it’s very unlikely that there’ll be no “noticeable skill” in both measures, as GD skill% is at least 74% for all seasons.

In 11 of the 15 years GF had a higher skill% than goals conceded and only once did it have a skill% <50%, 2003/04  Arsenal’s Invincibles season.

Overall we can take all 300 teams together and then figure out a better estimate of the “true skill%”. For each measure we get the following:

 

Totalskill3

 

So the amount of variance in goals scored due to luck in a season is slightly under 25%, for goals against it’s just under a third, and for goal difference luck makes up about one sixth of the total observed variation.

This means you would expect defence to be a less repeatable skill than offence, since more of the “variance from the mean” of a teams goals conceded is due to luck. Knowing this could be useful in prediction, as since goalscoring contains more skill it implies it should be weighted higher when forecasting results.

One thing I noticed is that GF has a noticeably higher skill% than points does over a season, I think there are three possible conclusions from this:

  1. Scoring goals does actually contain less “luck” than accumulating points.
  2. The poisson model doesn’t fit particularly well for number of goals scored, and the variance of goals expected due to luck is much higher. There is a noticeable covariance between the rate at which two teams score so this could potentially be affecting our results slightly. Although I’d like to think that the poisson model is going to be pretty close.
  3. The skill% Martin and James got were slightly too low. They suggested themselves that these figures were probably lower bounds, since considering home advantage and other effects could cause the percentage to rise. I did a quick check myself, using my data and home win% as 60% and got a skill% of 68%.

In conclusion having a good goal difference contains less luck that having a good goalscoring record, which in turn contains less luck that having a good defensive record. Pretty much as expected but it’s nice to confirm and quantify intuition.

Eventually I hope to run this analysis on leagues of differing quality, to figure out whether goalscoring consistently contains more skill than winning, even in the lower leagues.

StatsBomb Podcast: February 2016

Another edition of the StatsBomb podcast featuring James Yorke (@jair1970) and Benjamin Pugsley (@benjaminpugsley). Featuring:

  1. Title talk
  2. Fallen Giants: Liverpool, Man Utd and Chelsea
  3. Credit for Rooney, prodigal forwards and the rarity in which they maintain
  4. Postbag

Ben’s young forward charts can be found @objectivefooty [soundcloud url=”https://api.soundcloud.com/tracks/248341150″ params=”auto_play=false&hide_related=false&show_comments=true&show_user=true&show_reposts=false&visual=true” width=”100%” height=”450″ iframe=”true” /]   Also available on iTunes subscribe HERE   And if you enjoy, share widely. New work on the site today from @_PeteOwen : Luck vs Skill and Attack vs Defence Also coming soon:

  1. a Liverpool breakdown from Dustin Ward
  2. a round table opinion piece from a variety of contributors
  3. a MLS preview from Coleman Larned

 

The Fascinating Nature of Paco Jémez

  In the grand scheme of things in La Liga, there have only been three teams that have mattered this season; Barcelona and the two Madrid’s. Barcelona have become the Harlem Globetrotters, Real Madrid are their usual soap opera and Atletico are Atletico. You would certainly be forgiven if you took a glance at the rest of Spanish football and went elsewhere. But hidden under the jumbled mess below the top three is the continuation of a four year project that is one of the more fascinating you’ll see in Europe. Just by looking at GD and league positioning and you would be surprised why anyone would dedicate hundreds of words about Rayo Vallecano. 15th place and the 4th worst goal difference in the league. What’s so special about that? Thinking in that sense would be simplifying what Rayo Vallecano have achieved over the last few years. Spanish teams not named Barcelona, Real Madrid or to a lesser extent Atletico Madrid are already at a huge disadvantage but Rayo Vallecano take it to the extreme. Their annual budget is reportedly around €7-8M. Most Ligue 1 teams don’t even have that low a budget. The way that Rayo acquire their talent is mostly through loans or free transfers and with such a low budget, it leaves them privy to losing their best player on a yearly basis. The likes of Coke, Michu, Alberto Bueno and Leo Baptistao have left the club over the last 4-5 years. In a broad sense, there are two different ways to survive as a low budget team. You can go about it in the way like Angers in Ligue 1. You put all your eggs into a defensive system that suppresses shots at a crazy rate and hope that you can construct solid set pieces regularly to counteract deficiencies in open play. Or you can be like Wigan when Roberto Martinez was their manager, slant more towards a utopian style of football and hope that you’re not horrific enough defensively to survive. Rayo certainly slant towards the latter. Throughout his time with Rayo Vallecano,  Jémez has installed a concept known as Juego de Posicion. We don’t need to go through the gory details about it, so I’ll just leave this trusty link here if you want to examine it further. The best thing about Rayo Vallecano is no matter the opponent, their style of play doesn’t change. As a result of playing to win every time, very little Rayo Vallecano matches end in a draw. Over the last three completed season, Rayo have either had the least or second least draws in La Liga. This has led to a number of drubbings, including a 10-2 loss against Real Madrid and a 5-2 loss to Barcelona but both of those games involved Rayo taking an early lead through their proactive nature. https://www.youtube.com/watch?v=EVcXFvPHGC0 Despite the mad scientist way of thinking, Paco Jémez has managed to construct teams that are average to slightly above average in shot ratio. For a club that has such little in the way of money or assets this is a really impressive feat, especially when you consider a team like Valencia that have much more talented players are posting horrendous shot numbers. Rayo Of course not everything is peaches and cream. This season Rayo Vallecano are conceding more shots per game from inside the box than in previous seasons which is concerning. I’m not privy to expected goal data from previous seasons with Rayo but on this season alone, the defense has been pretty damn bad though not as bad as what the goal tally suggests. The decline in defense could be managerial issues or perhaps not having enough talent this season compared to previous ones to work the pressing style effectively. Something that’s pretty interesting to note with Paco Jémez has been his clubs not having save percentages that have approached league average. One could make the argument that the proactive nature of Rayo’s tactics can leave them exposed and concede high quality chances over and over again. Some of it might be valid. Only Levante have conceded more shots coming from counter attacks this season and Rayo rank in the bottom five in shots allowed coming from a danger zone pass. , two of the most advantageous ways to create high quality chances. Then again Marcelo Bielsa might be the most all or nothing manager you’ll find and in his two years at Athletic Bilbao, he had save percentages of 74.4% and 71.7% so maybe it could just be that Paco Jémez has been dealing with the bottom 10% of goalkeeping talent this entire time. An interesting comparison that could be made with Paco Jémez is Mauricio Pochettino. At least in a simple way they have both constructed sides that press the opponent and if we look at a basic statistical comparison, we can see that there is some similarities between Poch’s time at Espanyol and what Paco is currently doing with Rayo Vallecano. Paco vs Poch We know by now what’s happened to Mauricio Pochettino since leaving Espanyol: Southampton took a gamble on Poch, which was a really unpopular move at the time, and he could be managing a Premier League title winner in three months. Obviously this isn’t close to an exact science in terms of a comparison and I’m not privy to the exact details of how Espanyol played under Pochettino to make this comparison more apt, but I’d assume clubs could find ample footage of Espanyol matches under Pochettino and examine the differences/similarities much further than just shot numbers. This is probably sticking my neck out there but I think Paco Jémez is a really good managerial prospect. He’s produced very impressive shot numbers considering the context involved while in a style that’s almost mini Barcelona, enough of an impressive feat that it got rave reviews from Pep Guardiola. In comparison to someone like Mark Hughes who doesn’t seem to elevate underlying numbers when given quality talent, I’m much more confident that Paco is the opposite. Of course there’s risks with hiring such a proactive manager and you would like to think that with considerably better talent around, he’ll be a bit more pragmatic in terms of knowing when to risk the balance of creating high quality chances through pressing versus conceding them. There are a host of interesting medium-big clubs that could be on the hunt for a new manager this summer. Say Ronald Koeman decided he’s taken Southampton as far as he could and he’s sick and tired of his best players being poached up every summer window, Southampton could look at the success they had with Pochettino and think that Paco could achieve very similar things with a higher budget and a style that could be prone to greater positive variance. Other clubs likes Marseille, Roma, Lyon, Swansea would in their own ways fit the type of profile that Paco would want for his next gig. If the success of someone like Mauricio Pochettino is anything to go by, teams should be knocking on Paco Jémez’s door hoping that they could have one of the best up and coming managers be part of their future.

Predicting Shots and the Effect of Directness

A deep dive into Liverpool is coming soon but today we are talking shots.

It started with Benik Afobe. I find that’s true of the majority of my ideas nowadays. In this case, I wanted to know how many shots Bournemouth were likely to take against Crystal Palace so I could determine if Afobe was a good pick for my fantasy team. I wound up making a gut call but then came back around later to check if there was a way to easily estimate. There won’t be anything groundbreaking here and it’s possible or probable that most of you already mentally do this basically correctly: but it was interesting to see the data for me, so I’ll share.

 

Afobe’s goal was enough to see Bournemouth through and me into the money

 

The Wrong Way

The first thing that came to my mind was just averaging shots for from the attacking team against shots for from the defending team. I did this for each set of games (home and away) between teams in Ligue 1 and La Liga from last season. I then sorted the teams into high shots (top 5) and low shots (bottom 5) offenses and defenses. I then simply compared how many shots were actually taken in buckets of games involving these teams to what we’d expect with the simple average. As you can see, this simple average is a badly mistaken assumption.

 

Snip20160210_48

 

The graph shows how using this method consistently underestimated the amount of shots high shot teams took against high shot defenses (actually took ~15% more than rough average) and overestimated how many shots low shot teams would take against low shot defenses (actually took ~15% less than rough average).

So what was the problem? If you get a team that shoots more than average, going against a defense that allows more than average they feed off each other like loud-mouthed hecklers and Donald Trump. If you have an attack that shoots more than average (like Marseille’s 15.2 compared to Ligue 1 average of 11.5) and a defense that allows more than average (like Guingamp’s 12.6) you expect a total of [(3.7 + 1.1) + 11.5]*2. This is the difference from average for the attack and defense times 2 for the 2 games they play. So in this case you expect Marseille to take 33 shots total in the two games, they took 32. When you use this method, you get results that closely track with reality:

 

Snip20160216_67

 

Each bucket has an n of ~100 games. You can predict shot totals as a group pretty close to exactly using this method. None of the groups are more than 5% off the reality.

Why Is This?

I wanted to know the mechanism that allowed this sort of coupling effect. The most obvious place to start was completions per shot, a measure of how direct a teams attack and defense are. You can learn a lot about how a team attacks and defends checking this metric. Atletico Madrid’s fearsome defense last year forced opponents to complete 38 passes for every shot they allowed while Marseille’s chaotic high press acted as a sieve at (many) times, allowing a shot every 20 completions. Teams had to string together twice as many completions to get a shot vs Atleti as they did vs Bielsa’s Marseille or Paco Jemez’s Rayo. So I figured that this metric was what I should check first to explain the shot conundrum. It turned out some interesting results. When very direct teams play defenses that allow very direct attacks, things get positively linear. We also see when we have two slow or “molasses” teams, we wind up seeing attacks at an even slower pace than you would expect.

 

Snip20160216_68

 

“Predicted” means what you’d expect averaging the two. So when Espanyol’s direct attack (22.5 comps per shot) took on Rayo’s wide open defense (19.1) we expected 20.8 completions per shot. In reality Espanyol took 20 shots on only 335 completions (15.8 completions per shot). We do this for the top 5 most direct and least direct teams in each league to create the buckets. So sometimes teams can get more shots than you might expect from a certain amount of possession simply because the tendencies match up. They don’t have to work near as hard to get them.

Obviously, possession also plays a big part in how teams get shots and we aren’t dealing with that here at all. I’ve seen models that try to predict possession, and have dabbled in it a bit but never tested anything. If we can build an accurate model of who is going to have the ball and how they are going to work to get a shot off, that’s a good bit of the way toward modeling a game in a detailed way that I haven’t seen done too often. Of course, once someone scores you then have to re-adjust all those numbers based on the teams tendencies at different game states but that’s for another time.

What Does This Mean (for Leicester)?

It’s the time of year for special Leicester stats, and here’s my contribution.

 

Snip20160216_69

 

They are the most direct team in the league, which isn’t surprising.  We see United and Swansea on the other end, also unsurprising. So when Leicester plays Newcastle, Stoke, and West Ham (the most “direct” defenses) they are set up pretty well to get their shots with less work, so if they can keep their possession levels at expected levels they will get more shots. When Swansea faces United (the least “direct” defense, they have to work that much harder to get each shot off.

Does this knowledge help a manager?

Knowing how your opponent passes to get shots should be a minimum for any pregame scouting. Knowing how your own defense amplifies those tendencies is another piece of knowledge to keep in mind. If you are van Gaal and know Swansea is coming to town, you can pretty much count on your team being able to slow their attack to a crawl. Maybe this gives you a reason to pick a smarter, better in the air, ball-playing center back instead of the one with more pace. Maybe this means you tell your fullbacks to get forward quicker knowing Swansea’s buildup will allow them more time to retreat if caught upfield. Of course you have to weigh this with a lot of other factors and I’d assume many top managers have a good intuition for things like this already, but the information can’t hurt.

If you are Crystal Palace (the team that allows the quickest shots per pass) facing Leicester, you should be aware that you are even more at-risk to their lineup of direct sprinters, they are poised to strike even quicker than normal to get a shot off. Maybe you change how you approach corners or tell your players to play a few less risky passes.

None of this is groundbreaking, but it wasn’t something I went into games aware of before. Now I have a better pregame framework for which teams will be taking a lot of shots and how they will move the ball to take those shots. This wouldn’t have changed my thinking too much on the agonizing Afobe-or-not-Afobe decision but it’s good to know for the future.

Home vs Away Coda

I didn’t factor in home and away for any of the above calculations because I was looking at both legs in total. But, if you want to know a bit about the home/away breakdown: Home teams generally take 55% of shots in any given game. Home teams are also more likely to put those shots on target (34.1% to 33.6%) and score more of their shots on target (G/SOT of 31% compared to 29.5% for away teams). Home teams take shots from closer in (19.3 yards compared to 19.9 yards) while the only edge away teams get is a lower % of shots that are headers (12.5% compared to 14.1% for home teams, though even those come from slightly further out 10.3 to 10 yards).

 

Snip20160216_71

Data provided by OPTA

 

Super Sunday, Predictions And Palace: Premier League Round Up

Super Sunday Super Sunday came and went as a thrilling spectacle for the non-partisan bystander.  The two games featuring the top four contenders–who by any reasonable estimation have all but sealed their places–were full of contentious decisions, tension, storylines and thrills.  Even the most stone-hearted fan could surely not have been slightly warmed by the returning Danny Welbeck and his winning contribution and significantly, Arsenal’s late winner kept the four teams within a six point range.  Everything about Arsenal’s title challenge is reliant on squad fitness so from that perspective a deep bench for the first time since around November was a huge bonus. Manchester City may feel most aggrieved, specifically over the ludicrous penalty decision given against them, and now find themselves a small margin behind the other three. While it would be foolhardy to entirely discount them, they would need to be pretty faultless from here on in to seal the deal. It isn’t no chance, it has just decreased quickly and they have a match in each of the Champions League, FA Cup and League Cup final before they resume league duties. Thoughts may now well be elsewhere and not just in Pepland. For Tottenham, a creditable result against a backdrop of a concession of a high volume of shots. This is unusual and hasn’t been a problem for them since, well, the last game against City. In fact this was the first time they have conceded more than 13 in a game since the return fixture.  The difference between that game and this was here they continued their excellent record of limiting shots on target. Though Lloris made important contributions, he only made three saves and just once this season, again in the return fixture at White Hart Lane, has he made more than five in a game.  This is a consistent and impressive run and as much as their ability to generate attacking play underpins their run into the top slots.  They also haven’t conceded more than one goal against any of their rivals. Only Swansea, Stoke and Newcastle have managed two goals against them and nobody has managed three.  Their long stretch of having not been behind by two goals in a league game continues and consistency as much as quality has been a hallmark of this Tottenham side. Similar to Leicester, but unlike Man City and Arsenal, squad fitness has been good and continuing a theme from earlier in the year, the impact of a potentially decisive injury, that of Jan Vertonghen, has been minimised by the seamless integration of Kevin Wimmer, a player with a tiny amount of league experience. Of course narratives run strong and judging by a large amount of copy, we saw Leicester seal the title last week. Now we see them looking more vulnerable but with a clear schedule and tough fixtures behind them. The similarities between their game and Tottenham’s were clear, both took not undeserved leads via controversial penalties, but a soft red card raised the difficulty of their task to a level they could not repel. But only just. Small margins… Projections Out in the world of amateur statistical modelling, there have been many potential results forwarded regarding the most likely outcomes of this fascinating and topsy turvy Premier League season.  A former writer on here known to be well versed in betting markets noted this just yesterday: Can we just say that every single PL model projection for the winner is within the margin for error? This may sound blasphemous in a world that desires concrete predictions and answers, but is the most judicious takeaway right now. With only a two point spread and a huge range of factors likely to influence outcomes, many of which are unaccounted for by the modelling work done, it is simply unwise to declare a strong favourite.  Injuries are impossible to foresee as is fatigue and each could have a huge bearing, not only on this battle but on that for relegation too. Once more this weekend we saw the fine margins by which matches can be decided and the huge influence of fortune.  The team that eventually prevails will no doubt be one of the strongest contenders but also one of the most fortunate.  Each game this weekend had a selection of marginal decisions that could have been given or ignored. We also have a long history of teams defying expectation and a close battle brings into play extra factors: mentality, experience and other things statisticians find annoying to measure. Projections are a mere snapshot, a range of factors boiled down to often singular outputs where variation and bias can lurk unaccounted for and seemingly invisible. It can also be noted the set-up of the teams after 25 games in 2013-14; the two teams with the strongest attacks ended up facing off down the stretch and pivotal moments occurred right though to the end, margins are tight, winners and losers separated by little: 1314n Similarly just last season Leicester were seven points from safety, without a win in eight games and with two wins from 24 as late as 21st of March, yet survived and Sunderland’s near annual Houdini escapes have stung many a forecast and bettor. The many possibilities at both ends of the table are most often reflected by the somewhat relative caution found in the betting markets. This is not to denigrate the work done in private modelling, often people will be informing their own betting, but “skin in the game” counts for a lot. Omar Chaudhuri has done an excellent job of collating the Sporting Index spreads throughout the season and from that we can see that seven teams are now projected to be ten or more points out from their pre-season prediction, with Chelsea and Leicester up around the thirty point mark. That disparity for so many teams is a clear reflection of the non-usual nature of this season, with Tottenham’s dramatic and impossible to predict improvement, Watford’s solidity and Aston Villa, Manchester City and Manchester United’s underperformance all now feeding into the uncertainty. Prediction is never an easy game, the sheer volume of profitable bookmakers reflects this and this is a season in which formerly solid methodologies have struggled. To bastardise half a phrase heard in the week from Dean Oliver, “it’s okay to say we don’t know.” Pardew Puddled Crystal Palace have hit a wall. Without a win in nine matches, we find a familiar scenario for long time Pardew watchers. That is to say that his teams appear to be somewhat streaky, spells of excellent form are invariably followed by the opposite. Coming off a strong 2005-06 at West Ham, the following season he presided over “their worst run of defeats in over 70 years” (eight in all competitions), enough to find him jobless. Quickly rehired by Charlton he failed to save them from the drop. After a spell in the comparative wilderness he rolled up at Newcastle where he somehow rebuilt his reputation thanks to a turbo-charged 2011-12.  Subsequent seasons saw the rollercoaster dip once more culminating in a “total collapse” during late 2013-14 in which they went 5-1-13 in the second half of the season and lost seven of their last eight. He retained his job seemingly only through the belligerence of Mike Ashley. A return home to Crystal Palace has seen him helm a team that has outperformed shooting metrics for the majority of his tenure, but specifically this season. Seasoned Pardew viewers and those of us who track a team’s numbers have long expected some kind of reversion, and post Christmas, it appears to now have occurred. Now down in 13th, this seems far more in track with this team’s actual ability than the edges of the top six of seven they found themselves skirting pre-Christmas. What intrigues going forward is the question: will the rot stop? Indeed, is Alan Pardew capable of halting a downward slide? Away at West Brom and Sunderland are trickier fixtures than might have seemed earlier this season and then come three fixtures against Liverpool, Manchester United and Leicester. We have to go two further games down the line before the comfort of “home to Norwich” and by then it’s mid-April and without swift remedy, it could be looking quite dire. Relegation is extremely unlikely; with 32 points they should be more than safe with only a small handful of points-gaining games, but an inter-season reflection upon the success of the project could see Pardew vulnerable. It is probably a sufficient aim to maintain their position in the league, but any aspirations above that level may need to be tempered or adjusted going forward. Alan Pardew appears to effect limited influence over the success of his teams. OptaPro Forum Last week we found StatsBomb alumni old and new as well as some of the more well known and esteemed of the public football analytics blogosphere congregating in Central London for the third annual OptaPro forum. Also present were influential media contributors and many club insiders. This excellent event provided an opportunity for the groups to intermingle and communicate while enjoying presentations from a varied bunch of chosen parties. If any incentive was needed to inspire any StatsBomb readers towards data work, the story of Sam Jackson holds great promise. Having chosen to start a blog on the statistical analysis of goalkeepers only last July, he submitted to the forum and was understandably thrilled to be selected. Given the opportunity to present to a knowledgeable crowd, he was able to provide a new slant on goalkeeper work, and acquitted himself well. Often people are intimidated by a perception that advanced skills or complicated data is required to create new or innovative work and while there is no ceiling towards such analysis, there are still many truths to be ascertained from simpler more available data. Overall the skills and perception of the analyst can be the primary drivers and in advance of what one hopes will be a 4th annual OptaPro forum, now is an ideal time for anyone with an interest to get to work. As ever, we here at StatsBomb take a keen interest in the wider blogging community and endeavour to read as widely as possible across analytical work in all sports. It is never too late to get started, take an interest and look for truths in the numbers. Plus it’s good fun.       ____________________ Thanks for reading Follow me on twitter @jair1970        

Leicester And Liverpool: When Things Work Out And When They Don’t

When it goes your way… In losing to Leicester this weekend, Manchester City not only gave away crucial points to a title rival, they once again failed to record consecutive victories, a run that goes all the way back to October. This only weeks after handsomely leading the league by recording a string of victories without conceding.  Since their impressive start they have gone 9-5-6,a good deal less than title form and have failed to win more often than they have won.  Given their fantastic wealth and extremely talented squad, this is a disappointing return; not quite Chelsea, but nonetheless below expectation.  In advance of Guardiola they appear to have gone lame: injuries have bitten hard, with Navas, Nasri, De Bruyne, Mangala, Kompany and Bony all missing this game and a slew of kids on the bench.  Options were limited and the sight of Martin Demichelis facing the energy of Shinji Okazaki and raw pace of Jamie Vardy looked potentially troublesome, and so it was. Robert Huth continued his goalscoring form, Riyad Mahrez put another bat-signal up to Spain and it felt that we learned a lot more about City’s lack of youth integration and declining legs than anything else. For Leicester, when you’re a team having a season for the ages and your journeyman centre back scores three goals in two matches to secure away victories against top four rivals, it’s certain that things are going your way. Leicester spent the first half of this remarkable season winning matches featuring a ton of goals. Their defence wasn’t particularly effective at repelling shots or goals but high rates of shot conversion and the wonderful form of Vardy and Mahrez were enough to propel them towards the top of the league. Going into the tricky section of their fixture list, starting with Man Utd at the King Power back at the end of November, they were converting all their shots at a high rate of 15% which had effectively powered their 8-4-1  standing. During the tougher looking 12 games since–including two matches against Liverpool and Manchester City, plus games against Chelsea and Tottenham– they have converted shots at 14%; only a marginal decline. This has helped them go 7-4-1 but the biggest factor powering their run has been their save rate for in these games is that they only have conceded seven times from 40 shots on target, a crazily high rate of 82.5%. Spin that out to all shots and Leicester’s opposition since December have been converting under one in twenty against a usual league average of around one in ten.  All these rates tend to fluctuate through a season and rarely sustain for too long at either extremely high or low levels. For Leicester, so far, large parts of them have. When you base a season for the ages on year long super high conversion rates then ride a tough schedule with an extremely high save rate, it’s clear that things are going your way. For now Leicester’s 14.2% all shot conversion rate places them 4th in a list of teams during the era for which data is public (2009-10 onwards). The teams above need little introduction, each well known for attacking prowess that propelled title challenges: 2013-14 Manchester City and Liverpool and 2012-13 Manchester United. Two titles and a second place finish for these three, yet the following season found a second for City, a 6th for Liverpool and a 7th for Moyes’ Manchester United.  Leicester are in good, but unsustainable, company here. Each team brilliant in its own way, yet also at the absolute top end of conversion. And Vardy and Mahrez? One imagines legions of scouts bulking out non-league attendances looking for wayward teenagers with an eye for goal, or caravan tours around Ligue 2 grounds looking for players with a magic left foot.  That is a legacy will run long into the future and will likely have no yield.  Little to add to the last time I mentioned them, they have been brilliant, Mahrez especially; but to be an unheralded smaller team and find one break out player of the year candidate? That is a pleasant benefit to enjoy, but to have two? Yep, things are going your way. Regarding squad depth, there is still a huge gap between Mahrez and Vardy or indeed, Kante or Drinkwater and any replacements, one that long term will not be bridged by either of their sale or injury. The continued fitness of the key men in a comparatively small squad is of paramount importance, as it is for any team, but whoever replaces them is sub-par for Leicester’s current level.  Having maintained a steady team, and been free from injury, this is yet another aspect in which things have gone Leicester’s way. But what of the money clubs in this league? These teams are usually on the sharp end of title runs. Surely, they couldn’t all show vulnerabilities together, in the same season? Vast wealth has been spent by not only Manchester City but Chelsea, Arsenal, Manchester United, Liverpool and Newcastle. With such investment surely one or more of these teams could manage a sustained run at the title? Well…

  • Chelsea imploded like no other Champion before
  • Arsenal have suffered traditional injuries and undershot expected goal numbers by heroic volumes, see this from Will Gurpinar-Morgan for detail
  • Manchester United continue to be stuck in transition, pootling along in minor European positions
  • Weird chaos for Liverpool

When did this last happen? It’s been a while: the last time a team would have been five points clear of one of the traditional big clubs with a mere 53 points after 25 games was 2001-02. Newcastle had 49 points then and Manchester United and Arsenal 48. For Leicester to lead with 53 points is unusual too; go back to 2002-03 to find Arsenal on 53, three points clear of Manchester United.  Over the last ten seasons the average points taken after 25 games is 59.4, the average for second place is 54.1. For the league to be so compressed points-wise has become an unusual event, yet here we are; beyond rational prediction. Yet when things are going your way… Nothing here is intended to denigrate Leicester’s achievements, but their success will generate a ton of analysis attempting to look at the makings of a success story and a keenness from other aspirational clubs to mould a similar run. If Leicester were backing up this run with top notch shooting or expected goal numbers, there would be every reason to mimic them, but they aren’t. Is it likely that we shall reflect on this in years to come as the birth of “anti-taka” with half the league adapting their style by gluing a couple of Olympic sprinters onto an eight man defence? One suspects not. There has been much pleasure gained from Leicester’s ascent to the top of the table. It has provided a fascinating new twist to a league that had in recent years become largely predictable. The true praise for them can be reserved for Ranieri’s skill in improving them; not to their current status at the top, but from being a genuine lower half team last season to legitimate European contenders this.  For that is likely their true level; with this team, with fitness and a slight element of surprise, this team is above par in the league. Their shots and expected goal numbers peg them as lower tier European contenders, but if they fall to 3rd or 4th, to chastise them for failing to maintain a lead will be misguided; they will still have hit a very high mark. If they start next season and find themselves in the top half but a mile behind the top four, they will still have found a significantly improved level.  Liverpool in 2013-14 learned that special players can power long runs deep into the season, but they also learned that without them and given reversion over time, it can be very different. But when everything goes your way, it’s best to enjoy the ride before it stops. When everything doesn’t go your way… For Liverpool, it’s not even funny any more. Fans with other allegiances may disagree but the slump that Liverpool have found themselves mired in has been deep, repetitive and familiar. This weekend found a new low, with the law of sod taking full effect as they blew a two goal lead against one of the league’s worst teams, Sunderland, within minutes of a fan walkout. Sometimes, stupid stuff doesn’t go your way, like for example this:

All the while their head coach is suffering with a suspected bout of appendicitis? Soap operas would reject this serious of events for implausibility yet this and further defensive woes have been enough to undermine a reasonably encouraging numerical base.  A lot has been written about the low quality of shots during Jurgen Klopp’s reign, but regardless, they have taken a good volume of them: nearly 17 per game. The on target rate is comparatively poor at around 4.4 per game (26%), which reflects the lack of success from range but where the silliness occurs, where things haven’t gone Liverpool’s way, is at the other end. The positive side is that they have conceded very few shots. In twelve out of seventeen Klopp games, they have allowed nine or fewer shots. Only defeats against West Ham and Leicester have seen the opposition create a good volume of shots and shots on target and otherwise, for volume, they have been superb at suppressing the opposition. The trouble starts when you refer once again to how many of these shots have been kept out of goal. Season long, Liverpool have saved 57.6% of the on target shots they have faced compared to a league average of 69.4%. This is a lower rate than Bournemouth, who have had a horrific season preventing chances and goals and is lower than any full team season recorded over the last six years.  We can make it look worse by looking at Klopp’s era (56.0%) or taking it just to Anfield (48.5%).  No wonder home fans are frustrated; they are witnessing it right now and in front of their very eyes. While these figures are truly dismal, to endure it is also to this degree is unfortunate and extremely incongruous with the excellent repression of shots.  Despite the poor goalkeeping and individual defensive errors that have effected this misery, a rate this low just will not sustain in the long term.  Repression of opposition shots is an important trait to aspire towards as it is reflective of team quality, an inability to keep the ball out of the net is far more beholden to fate. The positive takeaway from all this is that some aspects of Klopp’s methods are working and project positively for the future. Some personnel tweaks in the summer, with obvious weaknesses up front and in goal, could well create some solidity where for now we see just mush.  Comparisons have been made with the aspects of Andre Villas Boas’ project at Tottenham: high shot volumes, suppression of opposition shooting and wonky conversion rates at both ends tick all those boxes. However, where Villas Boas’ sticky personality and stubbornness eventually saw his tenure fall apart, Klopp should be able to maintain enough charm to design a team capable of contending for top four next year. The last Liverpool season to get royally screwed by variance was 2012-13. They were an excellent shots team at both ends that year yet found a way to win only 16 games and finished 7th.  Of course that was the precursor to the Suarez-led rampage towards the title the following year, and while such ambition may seem out of range, this season the unexpected and unforeseen qualities of both Tottenham and Leicester were built from far lower bases. If they can do it then why not Liverpool? If only things could go their way… Ah hell. I warned against this kind of nonsense: using Leicester as a blueprint for another team’s future? An example of how each team should dare to dream and that anything is possible? That lazy comparison is something we may have to get used to.   __________ Thanks for reading! Follow me on Twitter: @jair1970

Stokelona: Where Myth meets Reality

  Everyone has bad memories of the Tony Pulis led Stoke teams. They were often rough, not easy on the eye and were prone to the occasional leg breaking injury. They did what they had to consolidate in the Premier League and in that sense, Pulis had a successful reign as manager. But these teams will never be revered or remembered for anything on the pitch beyond dirty challenges and goals coming from long throw ins. Even taking a look at the type of players Stoke brought in during Pulis’ last season in 2012-13 feels like harking back to a different era in football: Michael Kightly, Geoff Cameron, Charlie Adam and Maurice Edu to name a few. In the most literal sense, Stoke City have changed. One only has to look at their most recent signing in Gianelli Imbula. A dribbling, creative CM during Tony Pulis’ days would’ve had as much a chance of getting playing time as you or I. Now, Stoke broke their transfer record and spent £18.3M on him. Back in the day, the closest Stoke would ever come to having someone like Xherdan Shaqiri is if someone played manager mode on FIFA and bought him. Now they’re stocked up with silky (albeit flawed) midfielders: Bojan, Marko Arnautovic, Shaqiri, Ibrahim Affelay. Hell, they got a kid named Moha El Ouriachi from Barcelona. It’s pretty funny to see how much of a U-turn Stoke have taken when you remember the old days of Matthew Etherington, Rory Delap and Cameron Jerome. There’s no question that the transition to Stokelona has been aided nicely by the enormous TV deal. Go through every squad around midtable and you’ll find creative midfielders finding a home. Dusan Tadic (Southampton), Dimitri Payet (West Ham), Yohan Cabaye (Crystal Palace). Five years ago Stoke couldn’t offer the type of wages that were needed to attract the talent they’ve got with no guarantees of European football, but here we are now and Stoke have a trio consisting of Bojan/Shaqiri/Arnautovic. What a time to be alive! Flip through a newspaper (oh who am I kidding, no one reads those anymore) and you’ll probably find glowing praise for Mark Hughes. “Look what a great job Stoke have done”, “Stoke play attractive football”, “Chelsea should be going after Mark Hughes”. A very little bit of this is true. Certainly Stoke do not play the eye gauging levels of football that once was their staple and now rank in the top 10 in shots coming from throughball passes. That type of stat pre Hughes would have never come close to happening. Where the Stokelona mystique starts to go haywire is go past the admittedly fun tagline and the novelty; Stoke this season haven’t been a good attacking team, or even average. For all the potential silk, Stoke have concocted a below average attack. Whether using an xG model that value all shots or just shots that hit the target, Stoke are below average in shot quality and they rank in the bottom five in shot generation. Everyone remembers the 6-1 drubbing against Liverpool last year or the 2-0 wins against both Manchester clubs and the bonkers 4-3 win against Everton. Those type of performances put together have furthered the narrative that there’s a football renaissance going on (in ways, it’s similar to the hype around how Everton should be better than where they are). To drill this point even further, let’s play a quick game. I’ll give you two teams’ statistical profiles and you tell me which one is which:

Teams Goals For Shots Per game SoT per game xG for
Team A 24 10.8 3.3 26.87
Team B 23 9.8 3 24.48

All right, so who are the teams. Team A is 2015-16 Stoke City. Team B is… West Brom, managed by the angel himself Tony Pulis. Considering how everyone has utter contempt for the style of West Brom, having numbers that are within closing distance of them isn’t exactly flattering, so maybe the praise is slightly unfounded. Having said all that If you do squint hard enough, you do see a growing modernity to Stoke’s attack. Passages of play like this wouldn’t have happened under Pulis. giphy (18) giphy (20) The best attacking teams in football are able to throw multiple looks. Dortmund and Barcelona often alternate between attacks that are counter attacked based and others that are from established possession. Even in the PL, the likes of Tottenham, Arsenal and Man City at their best can overwhelm teams offensively by varying their offense. Stoke are nowhere close to that but you can see the makings of something interesting, particularly in that attacking midfield trio. What’s also been interesting has been the use of Ibrahim Afellay as at times a holding midfielder. Throughout his rather disappointing career, he’s been used as a winger but Hughes has experimented with him playing as a faux central midfielder. My guess for why this is the case is his ability to individually transition the ball from the middle third to opposition territory. giphy (19) Despite positive aspects to Stoke’s attack, there are massive problems. The obvious being that they don’t shoot enough. 10.7 shots per game is a paltry shot rate for a team widely perceived to play attractive football; even a peak Tony Pulis’ Stoke team in 2010-11 produced a considerably higher shot tally at 12.7. Last year, Stoke were a middle of the road team in terms of shot generation, ranking 9th with 13.2. Obviously not all shots are equal, and Stoke were below average in generating quality shots but it was an improvement over Hughes’ first season there. Another problem for Stoke is that there’s a statute of limitations on how good an attack could be without a decent striker. Liverpool are a good example of this whenever Roberto Firmino is played as a 9. Firmino does a lot of things you would want from a striker, especially in the context of what gegenpressing entails. Despite hitting a nice run of form recently, he just can’t do the last 10-20% needed that top strikers consistently can (generating tons of shots from good areas, finishing skill etc…). Bojan is in a similar boat. He has gotten some run as Stoke striker but he faces similar problems with Firmino. When Bojan has played as the 9, he’s averaged 2.5 shots per 90 which is fine given the context of Stoke’s shot producing problems, but in that role there’s a need for more. A tendency with Bojan is whenever he’s played up top, there will be numerous instances where one or both of Arnautovic/Shaqiri will make runs in front of him so he can either make long balls towards them or short throughballs. giphy (21) Bojan is just not a striker period. Neither are Arnautovic or Shaqiri. Arnautovic honestly might be the closest thing to a competent striker that Stoke have, which isn’t something you would hope for as an emerging club. Bojan at his best will buzz around like a little bee and create for others either with his dribbling or his passing. Him or Arnautovic are the best of a bad bunch. It’s been the cool thing in the media to give Mark Hughes loads of credit for what’s gone on. 11th in the table, two point away from 8th. Hey that must mean that Stoke are on the ascendancy. If Stoke do eventually become a perennial top 8 side, Hughes’ work will be looked upon favorably by the media which I somewhat understand on a basic level. However I do have some apprehension with the amount of praise that Mark Hughes has garnered for recreating Stoke’s attack. Yes, he’s changed them into something that’s considerably different from his predecessor and as a result, that’s helped in some small part to recruit better talent (again though, money talks and any PL club has more money than about 8-10 other clubs in Europe). But even taking those factors into account, there’s loads of evidence to suggest that Mark Hughes is more or less the same manager he’s always been no matter the talent or vision at hand. In comparison to someone like Pardew who has greater volatility, you’ll know what you get if you hire Hughes: middling shots for numbers and an overall midtable finish. Again though, I don’t want to totally slag Hughes as a manager because there’s something noble about his vision of recreating Stoke. Signing Imbula was a really smart use of asset managing and another sign that Hughes at least knows what he wants Stoke to be (even if his ability to actually take good players and make them fit into a healthy shot generating team has rarely been proven). Imbula’s playing style is like a younger Moussa Dembele and if he can ever find some end product, the world is his oyster. Even at such a young stage, he can be a considerably upgrade on Afellay when it comes to initiating counter attacks and defensively, he’s good enough to not rock the boat. Stoke are in the beginning phases of a revolution and there are some positives. They’re slowly but surely diversifying their attack and some of their recent signings show that the club is aware of what’s needed to compete for European spots. There’s a massive hole at striker which needs to be addressed. Mama Biram Diouf had an okay season last year but he’s 28 and for where Stoke are going, he’d be much better off as backup to whoever the upgrade is. I also think fullbacks are another place for upgrading soon. Glen Johnson at his peak was a very good attacking fullback but he’s been decaying for some time now. If the Arnautovic rumors are to be believed as well, Stoke will also have to be on the lookout for another attacking midfielder. The term Stokelona when it comes to Stoke’s new era of football can either be looked at in an endearing manner or a term purely based on banter. There are large issues at hand that need addressing both currently and in the near future, issues that make the current reputation of Stoke City football rather unreasonable. But hidden through all the narrative and hyperbole, there’s a project going on that’s very intriguing.

Atlético Madrid Without the Ball

The famous analyst Leo Tolstoy once eloquently stated “Every great attacking team is pretty much the same; every team that isn’t great at attacking is not-great in their own, unique way.” Powerful and flowing words. Why is this? Mainly because our statistical understanding of soccer is mainly shaped by the team with the ball.

We can measure most of what teams do with the ball and while 10 years from now we will look back on the rudimentary stats and conclusions we are reaching with amusement, we are least on a track that will lead us to a robust understanding of the game. When teams don’t have the ball we are still generally foraging in the dark. It’s not easy to get stats that correlate at even a .4 level while attacking stats correlate at .7 or more routinely. This makes not-great attacking teams often fuzzily look somewhat similar.

The one stat we know generally correlates well with preventing goals is possession. If you have the ball a lot, you can’t allow goals, think Manchester United here. This makes Atlético Madrid an even more fascinating case. Atlético are a, as Tolstoy would say, a “not-great” attacking team. They’ve scored fewer goals than Southampton, Eibar, Athletic, and Sampdoria. They don’t defend with possession at all: Las Palmas, Genoa, and Aston Villa are some of the teams who have had the ball more than Atleti’s 50/50 share.

Despite this, they are commonly floated as a UCL dark horse and led La Liga for a while purely on the back of their defense. They’ve allowed fewer goals than any team in Europe until this weekend, and even then hassled Barca at the Camp Nou with just 9 men. Arguably no team has a more unique defensive profile either, when I did a cluster analysis of every team’s defense in Europe last season the team hardest to find a group for was Atlético Madrid. We will take a look at the data here and see if we can find out more about how they play so well and so uniquely without the ball.

What do they do?
We will start with what they don’t do. We already noted they don’t keep you from scoring by holding the ball. When adjusting for amount of time the opposition has the ball: Atlético allowed fewer shots than anyone else in Europe last season. They allowed just over 10 shots per 400 completions (average was 15.4, with a SD of ~2). This year once again no team in Spain allows fewer shots per opposition pass than Atlético:

Snip20160126_16

They also don’t press high. Leverkusen, maybe the team most similar the Atlético in a philosophical sense, rely on heavy and high pressure to stop opponents. Atlético are basically league average in Spain when it comes to the high press.

What they do is to make it progressively harder for teams to move upfield. Of course every team does this, but Atlético stand out relative to any other team you can choose. We can see when opponents enter areas 70 yards from the Atlético goal, they are basically league average at stopping the pass. We see the z score increasing as teams get closer to goal, until they are basically 1.5 standard deviations tougher than average to complete a pass against when teams are getting into shooting areas.

Snip20160130_23

There is serious space between Atlético and 3 other top teams until about 30 yards from goal where Celta joins, then Barca and Celta eventually are harder to pass against from around 40 yards and out. Those are two of the highest pressing teams in Europe, so this should make intuitive sense. You can’t cover everywhere and Atlético are making the conscious decision to cover the most dangerous areas first.

Snip20160130_24

We can see that right around goal, no one in the league is tougher to complete against, in total. Above it says 2nd, those are entry passes only.

Snip20160126_17

Now plenty of teams decide to guard this most precious area of the pitch, but what often happens is the opposition simply has the ball too much in the area for the increased difficulty to make a big difference. If you are tougher to complete passes against, it doesn’t mean much if you are allowing other teams more opportunities. This is where Atleti differentiate themselves from other teams like Villarreal, Eibar, and Getafe who defend their area well is they don’t actually have to defend in front of goal that much. Last year Atleti opponents played a lower proportion of their passes in dangerous areas than any other team. This year the same thing is happening.

Snip20160131_28

Snip20160131_29

They don’t just sit back and guard the most dangerous areas, like Gladbach did under Favre. More than any other team in Europe Atleti make you work to even reach the area where you can start thinking about shooting. Once you get there your journey has just begun. Completing a pass is near impossible (toughest last two seasons), getting a quality shot is rough (2nd longest average shot allowed this year), and even trying to carry the ball is a slog. The average La Liga carry in the 25-yard danger radius is broken up after 10 yards.

This year Atleti break carries up after 6.2 yards. Last year, it was after 6.4 yards. These aren’t high-n samples but no La Liga team since 2012 has broken up carries quicker than either of their last two seasons. If you want to create a shot, you are unlikely to use a short, quick pass to get a quality chance. This is a map of all sub-20 yard passes that lead to chances created:

combine_images

Unsurprisingly, they allow the fewest amount of these short assisted passes as well. When you’ve finally worked so hard to claw your way into the danger zone and actually carry or complete your way into a (likely poor) position to get a shot off, no team blocks more shots than Atleti do (29%). They just never relent. To recap:

  1. Toughest to progress the ball into the danger zone against
  2. Toughest to actually complete a pass in the danger zone
  3. Toughest to carry the ball into the danger zone
  4. Toughest to play a short pass to set up a chance against
  5. 2nd (booo!) toughest to get a close shot against
  6. Toughest to get a shot clean through on goal against

So we know the what. Now can we figure out the how?

Hardworking midfielders

To pull this off you need a team that doesn’t stop working. Atlético’s midfield never stops working defensively. Tiago, Saul, Gabi, and Koke are the 4 players who have >700 minutes mainly in the midfield. Saul, Gabi and Koke all are in the top 10 in La Liga in tackles/90 by midfielders. Tiago, Saul and Gabi are in the top 11 in INTs/90. Tiago, Saul, and Gabi are in the top 5 in shots blocked by midfielders. None of these counting stats equals “Good Defense” but to see 3 different players at the top of all 3 defensive stats shows the effort put forth to stop the ball.

It’s not just midfielders: the team as a whole tackles more than any other team tracked on WhoScored (12 leagues). What can we learn from this? If you want to stop teams, you need probably 2 or 3 midfielders who can combine to form a wall and ideally they all go by one name. Building a midfield only isn’t enough, but it figures to be one of the key factors needed.

Is it reproducible?
I think it’s harder to reproduce than almost any other style. After getting past the point that defense is always harder to produce than offense, you can’t have any weak links. Atlético don’t and they deserve credit for that.

But who gets the credit here? It’s an important and difficult to answer question. Obviously we have to first give credit to the actual players working so hard and making the plays. Simeone and the backroom staff should get credit for molding these individual qualities into a wall. This doesn’t happen immediately.  The aforementioned players have been at Atlético for a long time now (Saul is the only one without >1000 minutes for 4 straight seasons) and this long working relationship has probably paid off in ways hard to replicate if you want to buy 3 players and plug them in today.

When you have 3 or 4 years with the same coach and same players with the right mix of personalities, it’s believable that you can accrue benefits from knowing where your partner will be, what they are thinking, and what they will do next. It makes sense to me that these “experience points” would be worth more on the defensive side of the ball, where teamwork among players is generally more important and tactical movements harder to drill. When you have the ball, it’s not that hard to know where to go as you can design movements when the ball is at a certain part of the pitch. When defending, it’s harder to be certain of what will be going on on the ball so the tactical possibilities multiply. Defenders with an understanding of each other could be helpful here.

Is it repeatable?
Yes. Last season Atlético was by far the toughest in Spain to complete a dangerous pass on and allowed by far the lowest proportion of total passes to come in the 25 yard radius in front of their goal. This season those are the two key metrics fueling their defense. Juventus has in the past shown an ability to repeat defensive metrics to a similar extent. This is more rare than elite attacks but happens enough to suspect they aren’t candidates for strong regression.

There is a downside to being so defensively-reliant. No matter how good your defense is, the general trend is a high-powered offense can set the game tempo against a defense. The variation for a top defense is much higher than a top offense and depends a lot on the quality of the opposition. When Atleti goes up against truly elite attacks, the edge still probably shifts slightly to the attacking team being able to create a few chances. They’ve just pushed the bar extremely high for the quality the opposition needs to be at to impose their will on the game. However, that will happen in the Champions League more often than it does in La Liga and makes them a candidate to be upset along the way by a team who probably isn’t as good.

Conclusion
We see teams who attack at an elite level all the time, across many different leagues. Atlético’s outlying numbers without the ball make them a special case. No other team can top their league in so many different categories and top Europe in many as well. It’s not as simple as packing everyone in your box and hoping your guys are stronger, to breakup play at a steadily increasing level without leaving gaps at the back requires commitment and quality from a large group of players and no weak links. It’s hard to imagine this lasting for too much longer or being commonly produced elsewhere, so for now we should enjoy one of the best teams without the ball we will ever see.

As always, data courtesy of OPTA.

Top 10 Postscripts Top 10 Lowest Proportion of Passes Allowed in DZ, 2014-15

  1. Atlético
  2. Nantes
  3. Napoli
  4. St Etienne
  5. Juventus
  6. Monaco
  7. Rennes
  8. Villarreal
  9. Lille
  10. Real Madrid

Top 10 Lowest Completion % Allowed, Intrabox Passes, 2014-15

  1. Gladbach, 26.4%
  2. Leverkusen, 27.9%
  3. Barcelona, 28.8%
  4. Atlético, 29.2%
  5. Villarreal, 31.1%
  6. Inter, 31.6%
  7. Bayern, 31.7%
  8. Torino, 31.7%
  9. Elche, 32%
  10. Lille, 32.5%