First lets rewind to the 28th of May 2011. Barcalona have just swept aside Manchester United 3-1 to win the Uefa Champions League. It’s domination. 68% ball possession, to go with 22 shots to United’s 4. Xavi completed almost double the amount of passes all of United’s central midfielders did combined and the quartet of he, Andres Iniesta, Sergio Busquets and Lionel Messi out passed the entire United team themselves. It’s Barcelona’s second Champions League crown in three years, to go with Spain’s World Cup triumph in 2010, and people from all places are once again praising the success of heavy possession football.
Forward the clock 11 months and Barcelona are in the Champions League semi final, their fourth out of four under Pep Guardiola. This time their dominance on the ball is even greater, obtaining the obscene figures of 79% possession in the first leg and 82% in the second leg. Only this time the result is not an emphatic victory. Chelsea win 3-2 on aggregate to go to Munich where they would win their first ever Champions League. A month later Guardiola would manage his last game for the club. Barca’s period as the standout best club in the world is effectively over.
The result was seen as a huge shock but we’ve seen an increasing number of instances since where sides with huge chunks of the ball have effectively been shut down by sides sitting deep and often ripped apart by fast breaks during the rare moments they don’t have the ball. First we had Bayern Munich’s destruction of Barcelona in the 2013 Champions League semi finals where despite a having much less of the ball they were the dominant force. The following April a changed Bayern under Guardiola were destroyed in a similar manner by Real Madrid in both legs, and in between Jose Mourinho’s Chelsea brought Liverpool’s 11 game winning streak to an abrupt end despite having less than a quarter of the possession. And in the World Cup Lois Van Gaal’s Netherlands, without the injured Kevin Strootman and with a back line that lacked the pedigree of some other teams, found solace in a counter attacking system led by the speed of Arjen Robben. The Netherlands’ three best performances in Brazil where the three times their possession was lowest.
But are such occurrences just infrequent results that will always come up when hundreds of matches are played per season? Where they just acts of chance in a sport where upsets occur far more frequently than most at almost 50% of the time? Or are they all examples of a recent trend of possession, an in particular extreme levels of possession, actually being detrimental to a sides chance of success?
Well, possession is certainly not a hindrance in a simple sense. In the Premier League, La Liga and the Bundesliga over the last three seasons there is still a reasonably strong correlation between amount of possession and the number of points a team gets in a single season. Bar Atletico Madrid twice, no team has got more than 65 points with under 50% possession, though Atletico did win the title in the second of those instances with 90 points.
However, can we confidently say that sides are having success because of possession? Or could it just be the case that better sides naturally achieve higher possession figures? Certainly you’d expect sides at the top of the table playing sides at the bottom of the table to dominate the ball. The better sides tend to have players who are naturally better at retaining possession due to superior use of space, ball control and passing. On top of this better sides usually have higher aspirations and need for a win, meaning they need to be constantly pushing forward with the ball. Whereas on the other hand, lower sides may be happy playing for a point, therefore willing to concede possession, with concentration on defence with the hope of the occasional break.
This is even true at the elite level, albeit it at a lesser extent. In the last three Champions League knockout rounds, sides winning have averaged 52.3% of the ball compared to 47.7 for the sides suffering a defeat. However, it is when we start to look at extreme levels of possession, where we see that possession can actually be detrimental to success, and that sides are becoming increasingly equipped at conceding possession and generating success.
Extreme possession at the elite level
As someone who has watched the champions league the last three seasons and developed into something of a possession sceptic, I wanted to divulge into whether possession superiority was beneficial or in fact detrimental. Specifically I wanted to look at whether possession can be a hindrance in extreme cases against opposition of a similar level. I didn’t want to look at marginal cases of possession superiority, as I don’t think small cases are either relevant to the pattens of the game or show any proof to which team is better at ball retention. And I wanted to compare sides of similar pedigree in order to eliminate the cases where possession superiority is the result of a large quality gap, or where a tactical decision to surrender possession is extremely unlikely to come off, due mainly to the talent gap on the field.
As my want to look into the matter has mainly developed from watching the Champions League in recent season, that seemed the obvious place to start. I decided to cull the group stages just because you still get large mismatches in the preliminary stages. (No offence intended to Viktoria Pizen, but their destruction at the Alianz Arena last season isn’t really that relevant as to whether extreme possession is a positive or negative for Bayern against Dortmund, Madrid or Chelsea.) I also set the parameters for extreme possession to be any instance a side achieved 65% or more.
From the 2012 Champions League semi finals onwards (where Chelsea beat Barcelona) there have been 18 instances where a side has achieved > 65% possession. In that time 5 have been won, 6 drawn and 7 lost.
To see if this was a trend that was consistent I decided to look at more matches so broadened the boundaries to any clash between top four sides in Europe’s to five leagues from the 2012/13 season onwards as well. For the current season I used matches between last seasons’s top four. I found 21 instances of extreme possession in league football, with 5 draws and 8 wins either way. Interestingly the results in different leagues are revealing. In the Premier League, Serie A and Ligue 1 there were only five instances, 3 of which were won by teams who’d severely surrendered possession. In the Bundesliga there had been 8 cases, we’ll call it the Bayern effect, with the > 65% team achieving 4 wins, 2 draws and 2 losses. La Liga also had 8 games, we’ll call it the Barca effect, but the arguable superior opposition (apologies to Jens Keller) caused 3 wins and 3 losses.
In total this left 39 games where there was a huge case of possession superiority, 11 of which were drawn, 13 won by the side with superior possession, and 15 won by the side without. So while the results don’t absolutely show that possession is a clear hindrance, they do show it’s not necessarily a positive either. We’ll have to divulge deeper.
Analysing shot data
Anyone familiar with football analytics will know that shot data is regarded as the best indicator of performance. Quantity of shots are one of the most repeatable statistics in the sport and, along with other factors such as the location of shots, are currently seen as the best indicator of future goals. As a result the number of shots a team takes, where they’re from and whether or not they’re on target are a great indicator of how many goals they’re going to score and the same applies for shots they concede being a measure of potential goals against.
In the 39 games were a team had > 65% possession the side who had dominated possession heavily outshot those who didn’t and in total had 645 shots at an average of 16.54 per game. Conversely they suffered only 335 at 8.82 per game. In other words, the side who dominated possession had a total shot ratio, the ratio of their shots to the total of shots in the game, (TSR) of 0.65, compared to a TSR of 0.35 for the sides that surrendered possession. To put things into perspective, Man City won the premier league last year with a TSR of 0.65 and Cardiff finished bottom with a TSR of 0.38. This is severe disparity. Yet the huge gap in shot quantity isn’t represented to the same extent when it comes to goals. The side with superior possession outscored the one which surrendered by a total of 67-50, which while significant, isn’t to the same extent that the gap in shot quantity was. A total goals ratio calculated the same way TSR is gives the superior possession side a ratio of 0.57 and the team which surrendered the ball a ratio of 0.43.
This can usually either be because of two things. First, the sides with the ball could simply have been unlucky. Or, their shot quality could be significantly poorer. We have to look more specifically at the respective shots that were taken. For all the 39 games I did my own shot location data recording, based somewhat on the excellent work of Colin Trainor and Michael Caley. I created three zones, primary locations, secondary locations and marginal locations. Primary would be all shots inline with the six yard box and inside the area, secondary shots would be shots in the box from wide positions and inline with the six yard box from deeper positions. Marginal shots would be all others, from deeper positions and poorer angles. It wasn’t particularly detailed, but would give a decent indication of where sides were taking their shots.
Somewhat surprisingly for me, the teams with less possession weren’t shooting from significantly better positions. My theory was that teams without the ball sitting deep would mean the side having the ball not being able to penetrate the edge of the box as much, therefore not being able to shoot as regularly from prime positions and requiring more hopeful shots from range. The sides with 35% or less averaged 38.81% of their shots from prime zones, compared to 35.19% for teams whoch had at least 65%. Possession sides had a slightly larger proportion of secondary shots but also had a similarly larger proportion of marginal shots. Sides with less possession shot from reasonably better positions, but the locations themselves were likely not significant enough to warrant such a contrast in the efficiency of the different extremes shots.
A deep wall of defenders, however, doesn’t only prevent a team getting into the box, it also tends to make shot quality from the same position worse. So far data relating to the location of defenders relative to on the ball actions hasn’t come about, so having lots of men behind the ball doesn’t effect ExpGoal calculations and the like, but the presence of nearby defenders and ones in line with the shot and goal are obviously going to make shot quality worse. Shots are more likely to be blocked and shots are more likely to be put off target due to less of a clear sight at goal the chance of a defender potentially intervening if a shot isn’t take quickly.
In the 39 games, the teams with greater position had 34.73% of their shots blocked, compared to just 21.79% for the teams which mostly played without the ball. The scatter graph below shows there is a correlation, albeit it quite a weak one, between amount of possession and the proportion of shots which are blocked.
Of course, a greater proportion of shots being blocked also means a smaller proportion of shots being on target. There is a inverse trend between amount of possession and the proportion of shots on target. The average proportion of shots on target of 30.54 in instances of extreme possession rises to 41.11 for the sides who have had less than 35% of the ball.
Indeed the TSR of 0.65 for the team with superior possession falls to 0.58 as a total shots on target ratio (TSoR) and vice-versa a change of TSR from 0.35 to a TSoR of 0.42 for the sides with less possession, which very closely matches the total goals ratios of 0.57 and 0.43 respectively. When you bare in mind that the teams which conceded possession where marginally shooting from better positions then it suddenly shows that they’re not being more clinical with their on target shots, or benefiting from a lucky PDO (the addition of a teams scoring percentage and save percentage) in a similar way, but are instead simply able to get a greater degree of their shots on goal due to the benefits of less crowded opposition thirds, which in itself derives from having less of the ball than the opposition.
The final thing I wanted to look at with regards to shots is the effect speed of attack has on the likelihood of a shot being converted. Colin Trainor discovered that speed of attack is very important when it comes to shot conversion rates and could potentially be even more important than shot locations. He discovered that the quickest attacks had the best conversion rates despite not having the highest proportion of shots in prime locations. While I don’t have speed of attack data at hand, it’s somewhat safe to assume that the side who spent considerably less time with the ball will have had faster attacks. Possession superiority usually leads to territorial superiority, meaning the side without the ball tend to shoot a lot more from counter attacks. In his research Colin also discovered that there is a strong negative correlation between speed of attacks and the average number of touches prior to a shot.
What all this tells us is that despite possession usually helping raw shot generation it doesn’t have the same help in scoring goals. This could be just a random trend, but looking at matches over the last 30 months, and nearly 1,000 shots in total it wouldn’t seem so, especially when there are legitimate explanations for such a trend, such as the location of shots, the number of defenders in close proximity and the speed of attacks. Indeed, when we only look at shots on target, there is almost no disparity between conversion rates.
With Bayern Munich and Barcelona sitting comfortably atop of arguably two of the three best leagues in the world at the moment it would be ridiculous to suggest possession football is dead. In most cases superior teams will dominate the ball and as a result the most successful sides will usually be among the sides who have the best possession figures. However what we can argue is that at the elite level extreme figures of possession have shown minimum signs of being a benefit and that certain factors have shown it to in fact be detrimental.
Higher levels of possession can help you shoot more, that we’ve seen, but it’s also much harder to get said shots on target. Arguably a high possession figure means a side has to do more in terms of overall shot generation than their opponent to counter act the better quality of shooting they concede due to suffering faster attacks, with less men behind the ball from slightly better positions.
Possession football may not be dead but equally conceding possession has never been more alive. Ever since Chelsea spent the best part of an hour with ten men, one proper centre back and just three defenders against the best side of this generation; conceded just once and scored twice with a fifth of the possession of their opponents, the world has seen that it’s not only possible to win doggedly despite rarely seeing the ball but also that it’s possible to destroy opponents with fast counters the way Bayern, Real Madrid and Holland have in the last couple of years. Top sides are more willing to concede the ball and hit sides on the break. People who use the term park the bus in a disparaging manner could be in for a painful few years as world football evolves and more and more sides become happy to use the tactic to their advantage.