Thursday, June 30, 2011

High Quality and Low Imbalance: Which League Has The Best of Both Worlds?

We're right in the middle of the summer transfer season. Lots of rumors swirling about, and scouts driven to distraction by the demands of coaches, agents, and executives to provide quick and reliable information about a player's performance and worth.

In a global market for players, many of the potential transfers involve the proposal to move a player from one league to another. So clubs want to be able to evaluate a player's performance in leagues as different as Brazil's Serie A, Germany's 2. Bundesliga, or the Eredivisie. Unfortunately, a basic question like "How much is a goal in the Eredivisie worth, compared to one in the Bundesliga or the Premiership?" is difficult to answer. Because football is a team sport, an individual's performance in the past has to be tied to the performance of a team (or even just a subset of the team) and in any given match. What's more, any prognostications about a player's future performance involves predictions about his play in and with the new team against a new and different set of opponents.

All this means that knowing that Falcao scored 16 goals in the Portuguese league last year or that Luis Suarez scored 35 for Ajax in 2009/10 doesn't tell us how many goals these players would score in La Liga or the Premiership. We suspect fewer, but how many fewer is the question that really matters. That depends on what we think makes someone score a lot or just a few in the first place, and this is where the devil is in the details. After all, there are at least three factors at work - the player's quality, the team he plays on, and the league he plays in (i.e., the teams he played against). Ignoring the first two for a second, let's take a look a league characteristics.

Monday, June 27, 2011

Scoring From Afar: Wigan's Average Goal Distance

(c) 2011

More On Wigan's Unusual Goals: A PS From ESPN

I've had lots of interesting reactions to my analysis of goal creation, and especially Wigan's unusual - as in "different from the rest" - pattern of scoring this last season. One of them was from Albert Larcada who works for ESPN’s Stats & Information Group, perhaps most prominently with ESPN's Soccer Power Index. Albert's work is always interesting, and he pointed me to a couple additional facts about Wigan's unusual pattern of play that appeared as part of an end-of-season review he did for ESPN's soccernet using OPTA's play-by-play data.

In particular, Albert found that Wigan’s non-penalty goals came from a longer distance than any other EPL team, and by a pretty decent margin. Take a look at this excerpt from his analysis.

Farthest Average Goal Distance (Player, Non-Penalty)

                     Team        Goals    Avg Distance (yds)
Charles N’Zogbia     Wigan         9           20.8
Didier Drogba        Chelsea      10           17.9
Nani                 Man Utd       9           17.3
Hugo Rodallega       Wigan         8           16.9
Wayne Rooney         Man Utd       8           16.1
>> Min: 5 non-penalty goals

This leaderboard is a bit harder to make than others as you must shoot and score from distance often. More than half of Wigan’s goals have come from N’Zogbia and Rodallega, leading us to...

Farthest Average Goal Distance (Team, Non-Penalty)
             Avg Distance (yds)
Wigan             18.5
Tottenham         15.2
Blackpool         14.7
Man City          14.4
Everton           14.4

Wigan’s average goal distance is more than a yard farther than any other club in the big 3 leagues (EPL, La Liga, Serie A). It’s not surprising Wigan was on the verge of relegation because of its inability to get easy goals this season.

So were Wigan innovative or so bad that they were unable to get close enough in front of the goal? Perhaps a young creative manager and a mediocre squad turned the Latics' campaign in 2010/11 into one of "When you have lemons, try to make lemonade!"

To read the rest of the article on, go here.

PS: These numbers were calculated with a few weeks left in the season so some of them may be slightly off from the final ones, but they're unlikely to be radically different when all is said and done.

Thursday, June 23, 2011

Has the Premier League Become More Unequal? Trends in Parity in European Football Leagues Since 1994

I've been traveling to Switzerland, Germany, and now England these last few days, so I thought it was only fitting to post the analysis below from London where I am currently enjoying the good weather and great company of friends and colleagues. When having conversations, especially here in England and over a pint or two of good beer, it's not unusual to hear complaints about leagues becoming too unequal. Usually these complaints are couched in terms of fairness, and they tend to come from supporters or officials of clubs that aren't doing as well as they would hope. But none of this means the complaints don't have merit.

To see if there's anything to them, some data Benjamin Leinwand and I have been collecting may help us out, I think. Initially, Benjamin and I wanted to get a sense of overall levels of parity in various leagues and to put them in perspective. So we started by examining levels of competitive balance in six European leagues with data for the past 15 years. We measured balance in a league based on wins and with the help of the Gini coefficient and the so-called Standard Deviation. The data showed that - considered over the entire 15-year period - the Dutch Eredivisie has been the most imbalanced league, while the French Ligue 1 has been the most balanced. The data also showed two groupings, with Ligue 1, the Bundesliga, and La Liga exhibiting more and the Eredivisie, EPL, and Serie A less balance.

Of course, there are different ways to measure competitive balance, and we'll provide some analyses of whether this matters before too long.* In the meantime, though, I thought I'd point out that what you find has a lot to do with where you look, and that's true when it comes to understanding league (im)balance, too. There are different ways of slicing the data on leagues and each gives you a slightly different vantage point on where leagues are or are headed. Here's an example. The good thing about combining 15 years of data when looking at league balance is that we can avoid focusing too much on recent events that may or may not be typical. But there's a downside to this: by averaging over a longer time span, trends get lost in the shuffle. It seems to me that both are important - we want to know if the Bundesliga is more balanced than, say, the EPL, but we don't want to make sweeping generalizations by looking just at last year's results.

But we also want to know if a league has become more or less balanced in recent years - that is, what has happened to parity over time. Is the Eredivisie's or the EPL's lack of balance (compared to the other leagues) a recent development, or does it reflect a long-standing pattern? To answer these kinds of questions and get a sense of the dynamics of competition in the six leagues, we  took a look at trends over time and plotted the annual Gini and Standard Deviation scores since the 1994/95 season. The picture looks like this.

The first point is that some leagues show very clear upward trends. The trend is most obvious in the case of the EPL, where we see an ever increasing imbalance in the league that started as far back as 1997. And this trend is apparent, regardless of which measure of imbalance we use.

Sunday, June 19, 2011

Why Wigan Were the Most Unusual Team of the Season: Goal Creation in the Premier League in 2010/11

Roberto Martinez has been in the news quite a bit lately, and for all the right reasons, it seems. His Latics managed to avoid relegation, and he was courted heavily by Aston Villa. He said no, apparently out of loyalty:
"Over the last two years the chairman has been very supportive to me and loyal, and now I feel I need to be loyal and supportive back to him. I haven't finished my job at Wigan Athletic; there is much work still to be done."
So what kind of work has Martinez been doing? Avoiding relegation is undoubtedly a major feat. As The Swiss Ramble, a blog on football finances, pointed out in an analysis of Wigan's financial situation:
"Given their substantial financial disadvantages, Wigan’s ability to survive in the Premier League is a minor modern miracle. They have the lowest revenue in the top tier, just about the smallest crowds, the highest reliance on television money, one of the highest wages to turnover ratios and no cash. As a recent set of accounts drily stated, 'The challenges on the pitch are very much reflected by the challenges off it.'"
Martinez seems to have been doing more than getting lucky, however. But what, exactly? One way to analyze Wigan's survival is to take a look at where their goals came from last season. To put things in perspective, a few days ago, I wrote about goal creation in the Premier League during the 2010/11 season. The upshot: for the league as a whole, most goals came from open play (about 70% or .93 goals per team and match), and the smallest number and proportion (2.6% or .036 per team/match) came from free kicks.

These overall totals are interesting because they give us a sense of where goals in top level professional football come from. Of course, this doesn't mean that all teams play the same style of football or create chances and goals in the same way, and it doesn't tell us whether Wigan (or anyone else for that matter) stood out. So how much variation in goal creation was there across individual teams last season? Can we recognize differences in team strategies by the way they managed to put the ball in the back of the net?

To answer the question, I went back to the data on shot and goal creation collected from the Opta/Guardian Chalkboards and calculated numbers and percentages of goals created from different match situations for each team. And lo and behold, they show that teams like Arsenal and Wigan created goals in very different ways. To see how much they differed and in what way, the next graph plots goal creation (the number of goals per match) from different match situations for each of the teams over the entire 2010/11 season.

You may be surprised to find out that the patterns of goal creation for each individual team follow the overall pattern of the league only in one important respect: all teams score more goals from open play than anything else. But once we go beyond this basic pattern, we see lots of differences in how teams created goals in 2010/11.

Thursday, June 16, 2011

Comparing the Competitiveness of European Football Leagues

Over the years, lots of people - from fans to managers - have wondered about competitive balance in football leagues. Is it healthy for the EPL to have teams like Chelsea or Manchester City with seemingly endless access to players, staff, and know-how? Or is something not quite right when the Wigans and Blackpools of this world struggle to not fall further behind? Does La Liga prosper when only two teams have a consistent shot at winning the league? Or is it better if a league can have 5 different clubs win the championship in the last 10 years, as we have seen in the Bundesliga?

Instinctively, most people would say that a football league that's too unequal will have trouble attracting fans in the long run. Lots has been written by academics in the field of sports economics about the upside of "competitive balance" - defined by the Financial Times in purely economic terms as a "market situation where no business is too big or has an unfair advantage." So how unequal is the Premier League really? And is it really more unequal than, say, the Bundesliga? And when do we know if a league has become too unequal? To answer these questions, we first need to figure out what we mean by unequal and how to put a number on it.

Tuesday, June 14, 2011

Why Shots From Free Kicks Are A Good Idea, Or At Least Not A Bad One

By Ian Graham*

I have been following Soccer By The Numbers for a while, and I usually love the articles Chris writes. I took a professional interest in his article on the Uselessness of Free Kicks in the Premier League: While we were developing the Castrol Rankings, we looked at free kicks and came to the conclusion that winning a free kick increased a team's chance of scoring a goal. I had also analysed shots from free kicks, and came to the conclusion they were no bad thing.

Chris' merciless demolition of direct free kick shots alarmed me: I had to look at the data again to convince myself I'd done the right thing. So, here's my defence of free kick shots.

Chris makes the point that FK shots are rare - it's a good point, but the really interesting question is what they're worth when you have the chance to perform one (after all penalties are rare, but clearly valuable). Chris showed that most teams failed to convert any direct FKs into goals, and that most of the teams that did manage it had poor conversion rates hovering around the 5% mark.

But we can't compare direct FKs to other types of shot - they come from outside the box, by definition! So I looked at the Opta data that goes into the Castrol Rankings to try and benchmark FKs against other shots more fairly. The data set includes all shots except penalties recorded by Opta from the Premier League 10/11 season.

The results? First I found that "normal" shots from outside the box (i.e. not direct FKs) originate about 23.5 metres from the goal line, compared to 24.5 metres for direct FKs (see graph 1). I think that's close enough to say we're comparing like with like. 

And when we do compare like with like, the true value of a direct FK is revealed! 4.5% of direct FKs are converted compared to only 2.6% of normal shots from outside the box. So, in an average match situation a player might nearly double his chances of scoring a goal by diving for a FK rather than taking a shot from open play.

I would say this is decent evidence that direct Free Kicks are a valuable resource. Phew. One final point of interest: Any sort of shot outside the box is clearly rubbish compared to one from inside the box.

* Ian Graham is the head of football research at Decision Technology, based in London. Since 2005 he has developed a set of statistical models for the prediction of football matches and the rating of players. At Decision Technology Ian delivers bespoke data analysis to European clubs, providing a framework for player recruitment and squad optimization decisions. He also works with the Premier League, consulting on a number of projects. He developed the Castrol Index, a novel player rating system used in both Euro 2008 and World Cup 2010. The Castrol Rankings project generates monthly player ratings for more than 2,000 players across Europe’s best five leagues. Since 2005, he has produced statistical research and Premier League predictions for the “Fink Tank” column in the Times.

Sunday, June 12, 2011

Where Did Premier League Goals Come From in 2010/11?

As you may know from reading this blog, I've been digging through some of the Opta/Guardian Chalkboards data on shot and goal creation in the Premiership and I finally had the time to take a quick look at some of the season totals to update the numbers I reported at the half-season mark.

A quick caveat: the numbers are subject to the definitions of match situations used by Opta/Guardian Chalkboards, of course. When comparing match situations, this sometimes makes things tricky and is worth keeping in mind (but obviously it doesn't stop me from doing the data analyses). It also means that we're sometimes comparing apples and oranges when it comes to match situations - more on that in a few days.

Also, a quick benchmark: there were a record total of 1,063 goals this season. This translated into 1.4 goals per team and match (1.398684 to be exact), and thus 2.8 goals per match. I'll take a look a trends in scoring before too long, so stay tuned for that. But in the meantime, I wanted to see how these 1.4 goals per team and 2.8 goals per match were created.

For the record, below are the numbers of goals created from different match situations in the Premier League in the 2010/11 season, according to the Opta/Guardian data. Drum roll please ....

Clearly, as was already the case by the end of December, the vast majority of goals in the 2010/11 season were created from open play - in fact, at a rate of .936 per team/match. The remainder (about .46 per team and match, since 1.4-.94=.46) were generated from corners (.186), penalties (.108), fast breaks (.083), and free kicks (.036). Another way to think about these numbers that it took teams 1.07 matches to score a goal from open play, 5.38 matches to score one from a corner, 9.26 matches to score a goal from a penalty, 12.05 to score from a fast break, and 27.78 matches to score a goal from a free kick.

Yet another way to slice these numbers is to look at what portion of the overall goal pie each type of goal constitutes. That is, how much did each match situation contribute to all goals scored? These numbers are shown below.

Thursday, June 9, 2011

Pictures of the Day: Revenues in the Big Leagues

Courtesy of the BBC News business section. Revenue-wise, it's not the Big 4, but the Premiership, the other Big 3, and Ligue 1.

And football's revenues keep climbing.

The most profitable league? Turns out that's the Bundesliga; also the league with the highest attendance. It also outpaced La Liga whose revenue growth is concentrated around a smaller number of clubs (Real and Barca). The Bundesliga's weak spot? Revenues from broadcasts. The least profitable league with the lowest revenue growth? Ligue 1; also the league with the biggest overall operating loss.

Tuesday, June 7, 2011

Goal Production In Six Leagues, Or: Are the Eredivisie and Ligue 1 Different?

Here's a (possibly) puzzling factoid of the day. Together with Benjamin Leinwand, I've been looking at goal production in different leagues over the last 15 years, give or take. And one of the (to us, at least) interesting and puzzling patterns has to do with how many goals are scored per match in the biggest European leagues - including the Bundesliga, La Liga, Serie A, the EPL, Ligue 1, and the Eredivisie. Since we've just started digging into the data, we thought we'd brainstorm publicly and share some of the things we see.

The data are totals by season and league. To provide a baseline, across the last 15 seasons and across the six leagues, the average number of goals scored per match is 2.68. During this time, the lowest number of goals was recorded in Ligue 1 in 2006/7 at 2.13; the highest was 3.27, achieved by the Eredivisie in 1998/99. Added up over the course of a season, that's a pretty distinct difference. And these minimum and maximum numbers for Ligue 1 and the Eredivisie do not seem to be an accident; as it turns out, these two leagues exhibit goal production patterns that are systematically different from the other leagues. Take a look for yourself.

Specifically, while Serie A, La Liga, and the EPL all have average goal totals of around 2.6 per match over the past 15 years (2.607, 2.652, and 2.603, respectively, to be precise), Ligue 1 and the Eredivisie look fundamentally different: the average for goals per match comes in at 2.331 for Ligue 1 but 3.023 in the Eredivisie. These are noticeable differences that beg for an explanation or at least are worth thinking about. So what is going on here?

Thursday, June 2, 2011

Comparing the Best Soccer Leagues in the World

This post is a slightly edited reprint of an article in Sports, Inc., a magazine edited by students at Cornell University in Ithaca, New York. To read the full issue, which is full of analytic insights about the major professional sports, click here or on the image below.

The article is a little bit of this, that, and everything from SBTN - some highlights and some low lights - intended for people with an interest in analytics but not necessarily expertise in soccer. One caveat up front: comparing leagues is inherently tricky business, if for no other reason that the data quality may be uneven and can be subject to limitations unknown to the analyst. As a result, differences attributed to style of play may reflect data collection techniques rather than real differences on the pitch. In any case, I bet regular readers of this blog will recognize most of it. Enjoy.