Europe’s Top Midfielders and the Uses of a Lamp-post

Most fans agree that United need a truly world-class attacking midfielder, who can consistently deliver the brilliant killer pass in the final third of the field. Someone like Sneijder, Ozil or (in our wildest dreams) Xavi or Iniesta.

Maybe we should start looking close to home. In our search for this missing piece in the jigsaw statistics might, as often, provide some surprising answers. When Opta lists the players who, on average, completed the highest number of successful passes in the final third of the pitch during the group stages of the Champions League this season the No 1 spot is held by Iniesta. No 2 is Xavi. No surprises so far. But in third place is – wait for it – Darron Gibson.

Un unsophisticated forecaster uses statistics as a drunken man uses lamp-posts – for support rather than illumination.

Who Deserve the Title? A Semiparametric Answer

In the Belgian premier league in 2008/2009 Sporting Anderlecht and Standard Liege completed the season with the same number of points. Although Anderlecht had the superior goal difference, a play-off match was organized to decide the championship and Standard Liege won it. But did they really deserve the title?

Gery Geenens at the University of Melbourne, Australia has taken a closer look at this crucial question and tried to investigate which team was really the best through statistical modeling,  using a semiparametric (don’t ask!) model. Among other things he points out that the semiparametric nature of the model grants it a great flexibility and allows the identification of previously ignored patterns. Maybe it’s now time to put this flexibility to good use and apply similar methods to ascertain whether United deserve to be in the top three of the PL at this moment.

Leaving aside the finer points of statistic modeling, it looked pretty obvious for about 80 minutes against Villa yesterday that we don’t deserve to be even in the top half of the table. Looking at the team-sheets you expected Villa to struggle in central midfield, but it turned out quite the opposite. For much of the second half Villa played us off the park. But judging from the last 15 minutes, we deserve to be champions. Why didn’t we see any of that drive, energy and penetration during the first 75 minutes?

So where does this leave us from a semiparametric point of view? Any readers with special insight into statistic modeling are most welcome to contribute!

Note to Fergie: Put Marc Albrighton on your shopping list.

I’ve only been able to find an abstract of Dr Geenens’ report, and I’m therefore  unable to reveal who actually deserved the Belgian title in 2008/2009.

“England at last have a coherent model of play”

Just one article for you to read today. Jonathan Wilson in the Guardian takes a long-term look at England’s lack of a natural left-winger and finds it might even be an advantage. He also explains how to get the best out of both Steven Gerrard and Wayne Rooney. In many ways the article is an Executive Summary of Wilson’s excellent book “Inverting the Pyramid.” Not to be missed.

The Best Team Unlikely to Win the World Cup (and so is England)

At the recent World Cup qualifier in Ukraine a poll among the English journalists present showed that one third believe that Brazil will eventually win the trophy, one third predict that England will lift the cup and one third think that someone else will. To most self-appointed analysts outside England (including this correspondent) this is a ridiculous exaggeration of England’s prospects. It now turns out that science has (kind of) proved that the sceptics are correct.

In a paper published in the October issue of Journal of Applied Statistics G. K. Skinner and G. H. Freeman conclude that the best team (which I would say is Spain, few would say it’s England) has a probability of only 28% of winning the cup, if it reaches the last 16. In reality the probability is lower since there is always a danger of being knocked out at the group stage. And the probability of England winning is again even lower, but the study doesn’t specify just how low.

The basis for their conclusions is the simple fact that the outcome of any football match is uncertain. The best team doesn’t always win. And in the knock-out stages you need to win four straight games to win the cup. In the latest World Cup only 5 games out of 64 had scores that indicated that you (or a gathering of statisticians) could be pretty certain (better than 90% confidence) that the best team won.

Are soccer matches badly designed experiments? by G. K. Skinner and G. H. Freeman (Journal of Applied Statistics).

 

Do we get more added time when we need it?

Following the controversy surrounding the amount of time added in the derby, the Guardian today publishes an interesting study that they claim shows that United benefit from an imbalance in the amount of stoppage time added to their matches.

The Guardian has looked at all of United’s league matches at Old Trafford since the start of the 2006-07 season and discovered that, on average, there has been over a minute extra added by referees when United do not have the lead after 90 minutes, compared to when they are in front. In 48 games when United were ahead, the average amount of stoppage time was 191.35 seconds. In 12 matches when United were drawing or losing there was an average of 257.17sec.

In 2006-07 United were winning 15 times on entering stoppage time and referees added an average 194.53sec. In the four games when United were not winning there was an average of 217.25sec. The following year the disparity was greater, Opta’s figures showing an average 178.29sec added when United were winning and 254.5sec when they were not. Last season it was 187.71sec compared to 258.6sec.

At first glance I’m unable to find any fundamental mistakes that the Guardian has made in this study. Could any of the, undoubtedly numerous, statistical experts among our readers help me prove that the newspaper’s findings are crap? Or do we actually get some preferential treatment at home?

A new way to beat the bookies?

A study published in the new issue of Journal of Quantitative Analysis in Sports tries to find a statistical measurement to predict the winners of international football tournaments. Ferda Halicioglu at the Yeditepe University in Istanbul argues that there exists a positive relationship between the competitive balance in domestic football leagues and success in international football.  Competitive balance is measured as the seasonal coefficients of variation (CV) of the end of season points. The CV lies between 0 and 1, with 0 indicating a perfectly balanced competition. A high degree of competitive balance means that the league members have relatively equal playing strength.

Halicioglu then goes on to rank the finalists of the Euro2008 tournament according to the competitiveness of the domestic leagues in the mid-term, i.e. the last six years.  The country with the most competitive league was France followed by Spain, Germany and Sweden. Well, Spain won the trophy beating Germany in the final, but both France and Sweden were eliminated at the group stage. The evidence looks somewhat inconclusive so far.

 Unfortunately the report, at least the version that I’ve managed to get hold of, is very vague when it comes to pinpointing the statistical evidence. It doesn’t get much further than to conclude that the analysis “suggests that the CV ranking could be used as a supplementary variable” to predict the winners.  Not enough to beat the bookies I’m afraid.

Professor reveals the perfect spot-kick

With the aid of high definition cameras behind the goals Professor Tim Cable at Liverpool John Moores University has studied all penalties in the CL last season and the PL this season to pinpoint the perfect technique for taking winning spot-kicks.

His findings are, well, less than sensational. To ensure a 100% scoring rate you need to follow these guidelines:

-The ball needs to cross the goal line at exactly 0,5m below the crossbar and 0,5m inside the post.

-The ball needs to be kicked at a speed greater than 65mph.

-This requires a run-up of 5/6 steps.

-The striker needs to commence his run-up from the edge of the 18-yard line with an angle of approach of 20-30 degrees to the ball.

If I understand this correctly, the professor has found that you’re likely to score if you hit the ball hard into the top corner. Which makes sense, I suppose. But I don’t comprehend why he needed high definition cameras to find this out.

But professor Cable certainly has some sound advice for penalty takers looking to improve: practice.

A brief guide to when to commit a red card offence

Defenders often face the decision whether to deny the opponents a scoring opportunity if this means receiving a red card. From now on Ferdinand, Vidic et al can rely on academic research to help them in this decision-making process.

In a study published in the latest issue of Journal of Quantitative Analysis in Sports three researchers from Colombia University, New York (Vecer, Kopriva, Ichiba) analyse the effects of red cards in soccer. One of the conclusions is that when the opponent’s chance of scoring in a particular situation is 57,5% or higher it’s better to prevent such a scoring opportunity from the very beginning of the game, provided that the offence doesn’t lead to a penalty but just a red card.

The tricky part might be to in an instant calculate whether the probability of scoring is above or below 57,5%.

Review: Myths and Facts about Football

This book draws together academic research from around the world, exploring a number of wellknown “facts” about football and submitting them to scientific and mathematical tests. Some are found to be myths, such as “teams run a greater risk of conceding just after scoring” and “taking the lead just before half-time makes a win much more likely”. While others are found to be true, e.g.  “goalkeepers dive too often for penalties” and “teams who celebrate goals collectively achieve better results”.

More than 40 scientists contribute to 20 chapters covering areas both on and off the pitch such as strategic choice, team behaviour, referee behaviour, demand, expert predictions, labour market and stock market. With so many contributors involved I suppose it’s more or less inevitable that the standard of the findings varies quite a lot. In previous posts I’ve touched upon post-scoring behaviour and the chances of scoring a penalty. Other interesting chapters deal with second leg home advantage and how accurately markets could predict the outcome of the Euro 2000 and the 2002 World Cup.

Some of the other chapters I find rather less interesting. In one case research even seem to take a step backwards when a study of stock market reactions to match results fail to use betting odds to control for expectations, unlike in previous research by Dobson and Goddard. This seems rather important as only deviations from expectations should, in theory, affect the stock market.

Nevertheless, this book is well worth a read, but I got a bit tired of it towards the end. It’s probably best enjoyed in small doses.

Myths and Facts about Football: The Economics and Psychology of the World’s Greatest Sport, Editors: Andersson, Ayton, Schmidt. Cambridge Scholars Publishing.

No football debate will ever be the same again

Today I take the unususal step of recommending that you spend your day at a website other than Between the Lines, as the Guardian launches its interactive chalkboards. They make it possible to track and analyse just about every event in every PL-match within 30 minutes of the final whistle. Every page should carry an addiction warning. Highly recommended!

But you have to question the motives of the Guardian. Are they trying to get us all sacked?