Thursday, September 29, 2011

Efficiency Ratios: The Premier League Does It Differently

The blog has been quiet for a few days, and for a reason. I´ve been busy working on a little side project. Before too long, I´ll tell you more about that, but one of the things that have come of out of it has been a more detailed look at various performance indicators across leagues with better and more recent data. Without revealing too much, I thought the next graph was worth sharing, especially since I´ve been writing about efficiency ratios of teams in the big leagues of professional football in previous posts.

One of the more interesting facts - well, at least to me - has been that leagues produce remarkably similar overall yields (goals to shots ratios), right around .111 or 1 goal in 9 shots. These yields don´t vary all that much across years, suggesting that there is an inherent stability in this efficiency metric and one that hasn´t really changed very much, if at all, since the early days of notational analysis in football.

At the same time, it´s becoming ever more apparent that leagues produce this basic overall goal to shot ratio in very different ways. How different? Take a look at the graph below, which shows the overall yield (in green), along with accuracy (accurate shots to all shots) and conversion (goals to accurate shots) rates, based on data from individual matches collected over the past five seasons.*

As the graph shows, La Liga and Serie A are most similar to one another, combining high levels of conversion with low levels of accuracy efficiency. This means fewer accurate shots, but those that are accurate are more likely to find the back of the net. In contrast, Bundesliga shooters produce accurate shots more efficiently and are about as good when it comes to converting accurate shots to goals. The league that´s most different from the others is the Premiership. Efficiency metrics in England differ significantly, revealing very high levels of accuracy but also much lower levels of conversion efficiency. Premier League teams get to the same overall yield as teams in the other leagues by shooting more efficiently and scoring less efficiently. Why this is the case is anyone´s guess - and I invite (informed and well-intentioned) speculation in the comments! - but it does suggest that Premier League offenses and defenses play a somewhat different game than teams on the Continent.

* Recall that accuracy * conversion = yield.

Tuesday, September 20, 2011

Graphs of the Day: Efficiency Ratios in the Top 4 Leagues for 2010/11

Haven't had much time these past few days to do much in terms of detailed analyses, so here's just a couple of fun (ok, "fun for geeks") graphs of efficiency ratios (goals to shots, accuracy, and conversion) for teams in the big 4 leagues during the 2010/11 season. Goals to shots is the overall yield. Accuracy is defined as the ratio of accurate shots to all shots created; conversion is the ratio of goals to accurate shots. No commentary from me, but comments always welcome, of course!

Next are the efficiency ratios - accuracy and conversion. The teams are ordered by the overall yields shown in the first graph.

Thursday, September 15, 2011

Cheaters Never Prosper? The Connection Between Fouls and Points in the Premier League

After Wigan lost to Manchester City last March, Wigan manager Roberto Martinez told the press that City stopped Wigan's momentum with fouls: "City used their experience," the Wigan manager said. "There were a lot of technical fouls every time we got into attack situations. Micah Richards should have seen a straight red for his tackle on Tom Cleverley, but the overall number of fouls shows how the game went."

Looking at the facts of the match, it's hard to quibble with Martinez's assessment. For much of the match Wigan were the only side with any ideas, and the Latics had more shots, corners, and possession in the match. City also committed more than twice as many fouls as Wigan (17 v. 8). But is Martinez right? Can teams gain points when they foul? What, if anything, is the correlation between fouls and performance indicators? To answer the question with some degree of reliability, I went back to the last five seasons of the Premier League and examined individual match data. In particular, I calculated the correlation between the number of fouls teams committed and the number of points they won. Here's what you see when you graph that statistical relationship for the league as a whole for each of the last five seasons. The yellow line shows the best fit to the data; the grey shaded area tells us that 95 out of 100 times, the correlation will fall into the range shown.

The graph shows a couple of interesting patterns. First, in each of the five seasons, more fouls are associated with fewer points won. But which way does the causal arrow run? Do bad teams foul more or does fouling more lead to fewer points?

The relationships between fouls and points or fouls committed and suffered are undoubtedly complex. From earlier analyses we know that less successful teams on average commit slightly more fouls. At the same time, bad teams don't always foul more; more importantly, while less successful teams may foul slightly more, this doesn't mean that more successful teams are fouled more. In fact, Martinez would argue it was just the opposite for his team playing Manchester City. 

So what we can say is that there clearly is a correlation, and it's negative for the league as a whole.* So teams that foul more win fewer points and teams that win fewer points foul more. Fouling seems a costly strategy, especially considering that fouls tend to beget fouls from the other side.

The second and perhaps more interesting pattern in the graphs is that the connection between fouls committed and points earned looks very similar for four of the five seasons. From 2006/2007 to 2009/10, the statistical relationships look very much alike. But the slope of the line is significantly shallower for the 2010/11 season. The obvious question is: why?

Tuesday, September 13, 2011

Fortress Bernabeu? Real Madrid Can Win the League By Playing Better Defense (Or By Barcelona Playing Worse Defense)

Not imposing enough?
The other day, Jose Mourinho made news by claiming that last season was his best ever as a coach. Truth is, he did do very well. Real came in second in La Liga, won the cup, and reached the semi-finals of the Champions League. Now, whether that's really better than any of the seasons with Porto, Chelsea, or Inter is another question. But there was one statement that got us thinking. Speaking about the La Liga campaign, Mourinho was quoted as follows: "It was the championship in which I won more points, won more games and scored more goals. It was a very, very, very good championship."

Looked at in this way, it's hard to disagree. It was a very, very, very good championship for Real - offensively speaking. While Barça is usually praised for their excellent attacking play, last season Real Madrid found the back of the net even more frequently (and lest we forget, Ronaldo won the Pichichi Trophy).

The numbers tell the story. For example, Real Madrid was superior in terms of shot production. On average, they produced 19 shots per match, compared to Barça's 15.5. Of course, the overall shot count is not entirely conclusive. It's also about the ability to fire a shot on target (and to score, obviously). For Real Madrid, shooting on target seems to be somewhat harder than for Barça. Although they take more shots, Mourinho's team managed to place about 42% of their shots between the woodwork. In contrast, Barça did slightly better, with a target/shot percentage of 46%. At the end of the day, Real scored 2.68 times per match, while Barça managed 2.5 goals. Given that Real Madrid needed seven shots on target to score once and Barça only six, the Catalans were slightly more efficient and Real compensated by shooting more frequently.

What Mourinho doesn't mention is that Barça had the superior defense: Iker Casillas let twice as many goals slip trough as Victor Valdés.

So here's our unsolicited advice, based on a few numbers: If Real wants to win the championship this year, they have to find a way to defend more effectively. We know: it might sound a bit odd to say that Real Madrid - much criticized for Mourinho's catenaccio style of play - needs to defend better. So let's take a look at the numbers to see what they tell us.

Thursday, September 8, 2011

Defensive Efficiency in the Premier League in 2010/11

A few days ago, I wrote about different ways of measuring a team's performance - either in absolute or relative terms. Relative metrics are useful if we are interested in drilling a little deeper into what players or teams are able to do, given the resources they have or in relation to some other event on the pitch, not simply whether they do a lot or a little (like scoring or possessing the ball). In those analyses, I looked at a team’s offensive productivity in terms of its ability to score goals efficiently. A team is more efficient offensively if it achieves goals with less effort or expense - in that particular case measured as the ratio of goals to shots taken.

The logic of efficiency should also apply to teams' defensive performance. We can investigate how well teams defend by examining their ability to prevent shots from finding the back of the net - that is, by calculating the ratio of goals to shots conceded in each match. Take a look at last season's defensive efficiency figures for the Premier League (by club) in the graph below.

As in the case of offensive efficiency, some of the best teams are found at or near the top.

Sunday, September 4, 2011

Efficiently Offensive: Premier League Teams in 2010/10

Tomorrow is Labor Day here in the United States. So naturally, I've been thinking about how we measure work performance in football. It's also a long weekend, with most people off from work for three days. If they spend it catching up on the Premier League season so far, they could be forgiven for scratching their heads upon finding out that last year's top scorers in the Premier League Dimitar Berbatov and Carlos Tevez have yet to start a singe match this season. Why would teams handicap themselves by having these prolific scorers come to work but then leaving these potent weapons on the bench?

(c) Gordon Flood
The question is reasonable. But it also implies that more of something is the best way to measure what footballers do at work. Talking about team performance in terms of how often things happen seems natural. "How many goals did he score?" or "how many shots did they concede?" are questions we might ask about a player or club. Looking at performance in this way means analyzing the frequencies of events - the "how much" of football we see. It's a way to gauge performance in an absolute sense - who did the most of something (like scoring goals)? That's useful as far as it goes. But there's another set of performance metrics that can be very informative. These measure performance in relative terms. The most common is a percentage or ratio - think pass completion percentage, goalkeeper save percentage, or the goals to shots ratio.