One of the first football stats to break through to the big audience has been the teams’ amount of possession of the ball. At present virtually every televised game is, at some stage, illustrated with stats concerning possession, which makes the general audience, at the very least, aware of the difference in this regard between teams. Traditionally, more possession has always been related to better performances. “As long as we’ve got the ball, they can’t score a goal” is one of many famous quotes along similar lines.
Possession and performance
Recent data, however, published for MLS matches as well as recent Premier League data, shine a different light on the relation between possession and performance. The issue of how to measure possession is one area of debate, while the relationship between the measured time of possession of the ball and performance in terms of scoring goals and winning matches is quite a different one. Aside from that, there’s the intriguing issue of how to measure possession, which has been debated on this website before.
Virtually the only publicly available possession stat is presented as cumulative time in possession of the ball, usually expressed as a percentage, split between both teams. Intuitively, one would think that a positive relation between time of possession and the amount of goals scored would exist. But, as recent data both on the MLS and the Premier League quite clearly show, such a relationship does not exist.
This can be explained by the fact that not all time in possession of the ball provides equal goal scoring opportunity per minute of possession. Short bursts of possession, more commonly known as a counter attacking game, have a better chance of scoring goals per minute of possession, than longer spells of possession do. So teams outperforming their opponents may have more time on the ball, reflected in higher possession stats, but their opponents have a higher quality of possession. Somewhere in the middle between this quantity (time of possession) and quality (goals per minute of possession,) teams should ideally find their optimum share of possession.
A further word of caution when looking at the relation between possession and performances needs to be addressed at the direction of the relation. Simply said, teams defending a lead may concede possession, while teams chasing a lead may adjust their tactics in order to chase an equalizing goal, thereby increasing their share of possession. The finesses of such a relationship still need to be explored in more detail.
That being said, at present the only data available for all Eredivisie matches is possession in terms of the ‘old-fashioned’ time parameter, again expressed as a relative proportion between both teams. Ranking the Eredivisie clubs according to their average share of possession, with data provided by Infostrada Sports, during the first 11 match rounds of the 2011/12 Eredivisie provides the following picture.
Ajax tops this list with an eleven match average possession of 57.6%, while only the top- and bottom four teams, in terms of possession, find themselves outside the 47%-53% zone. Most higher quality teams find themselves in the higher ranks of this list too, but there is no clear correlation with performance at first sight. Overall, the average winning team had a possession of 53.3%, with losing teams having the remaining 46.7% share of possession.
Things get more interesting if we calculate a parameter that was recently introduced on the excellent 5addedminutes blog by Omar Chaudhuri. He described the relation between possession and performance by calculating the average time of possession per goal scored, and reversely, the average amount of opposing team possession per goal conceded.
If we take league leaders AZ as an example, we see that they’ve had an average share of possession of 52.9%. During their possession of the ball, they managed to score a goal on average every 21.5 minutes, which would ranked them fifth in terms of converting possession into goals. However, AZ’s excellent start to the season is much more related to their defensive performance, compared to their offense. On average, AZ conceded one goal per 68.4 minutes of opponent possession.
This metric allows us to divide the Eredivisie teams into three groups. Teams whose offense outperforms their defense, teams whose offense roughly matches their defensive input, and finally teams like AZ, whose defense outperforms their offense.
The best example of a team whose offense performs better than its defense is Ajax. They rank third in terms of scoring goals (once every 20.4 minutes of possession), yet only tenth in terms of preventing goals (one conceded per 30.0 minutes of opponent possession). Further examples are Heerenveen, NAC and De Graafschap
A fine example of a team leaning on its defensive qualities, besides AZ, seems to be Vitesse. The Georgian owned club occupies the second place in the defensive chart, conceding only once every 50.8 minutes of opponent possession, yet they are ranked tenth in terms of converting their possession into goals, with a goal per 36.0 minutes of possession. Further examples of teams whose defense outperforms their offense are ADO and N.E.C.
The remaining group consists of teams who put in a rather balanced performance in terms of defense and offense. Twente, who lead the offense charts, is an example of such type of team, presenting a third ranked defensive effort to go with their offensive qualities. Further examples are PSV, Groningen, Feyenoord, Heracles, Utrecht, RKC, VVV, Excelsior and Roda.
This post could never have been created without the support of Infostrada Sports, who’ve provided the data for the analysis. Furthermore, a word of thanks goes out to Omar Chaudhuri, for launching this post, introducing the idea of calculating goals scored and conceded per minute of possession.