Utilising possession in the Premier League
A number of questions have been raised recently over the usefulness of possession statistics in football. Liverpool’s Director of Football Damien Comolli claims that “there is less of a correlation between possession and success than we supposed… it tends to be overrated by most clubs.”
Both Soccer Statistically and Chimu Solutions (here and here) have looked at data from the MLS, and found that teams are actually more successful in matches where they see less than 50% of the ball, a counter-intuitive phenomenon.
They offer various explanations as to why this might be the case; bearing in mind the direction of causality (do teams sit back on a lead?).
Last season’s Premier League possession data, provided by EPL Index, shows a similar trend, to the extent that possession is a poor indicator of a match result. Winning teams averaged 50.1% possession; by extension losing teams’ average possession was 0.2% lower.
Statistically, these values are no different, and whilst there’s nothing to suggest that teams benefit from having less of the ball, there’s equally nothing to support the age-old belief that dominating possession is crucial to success. (Football365’s Mediawatch neatly tackles the Daily Mail’s Martin Samuel on this point.)
It’s perhaps a case of not how much you have, but how you use it. Based on average possession numbers from last season, Arsenal appear to be the immediate culprits in this respect.
Minutes in possession per goal scored
Yet surprisingly, only Manchester United were better than the Gunners last season in turning their possession into goals, taking on average just under 29 minutes on the ball to score (i.e. possession percentage multiplied by total minutes, all over goals scored). This may serve as vindication of Wenger’s philosophy on the pitch, though as we will see the main concerns are at the back.
It’s no surprise that Stoke are third on the list; Tony Pulis’ men need to be ruthlessly efficient with the ball given they are without it for prolonged periods in most matches. Meanwhile relegation for Birmingham was a culmination of enjoying only 47.6% possession on average, and failing to make good use of this limited possession. In contrast, West Brom were productive with their time on the ball, though perhaps more intriguing would be a Di Matteo-Hodgson comparison.
Minutes out of possession per goal conceded
From a defensive point of view, the focus shifts to Man City. Not only did Roberto Mancini’s men retain a substantial amount of the ball across the season, they were superbly resolute without it, conceding every 49 minutes on average when out of possession. Mancini’s faith in his defence has clearly formed the base for a more enterprising approach this season.
Stoke again come out favourably, needing and proving to be defensively stubborn in the face of consistent pressure. A 3-1 home win against Arsenal characterised this remarkable efficiency, conceding once despite their opponents enjoying nearly 67 minutes of possession in total, and scoring a goal just under every 8 minutes when in possession themselves. Needless to say the match also highlighted Arsenal’s frailties without the ball.
In comparison to the final league table, there’s a stronger correlation between points and minutes out of possession per goal conceded than with minutes in possession per goal scored, providing yet more evidence that, given a choice, teams should focus more on their defensive duties than attacking plans.
Do these statistics support a case that Stoke should seek to retain more possession, given their efficiency at both ends of the field? Surely if they chose to keep hold of the ball for longer they would score more and concede fewer? The answer is probably no; one of the reasons they are so strong defensively is perhaps because they have so little possession, as it enables the team to retain their shape and organisation at the back.
These statistics are useful in noting how well teams utilise time on and off the ball, and may be good compliments to looking simply at goals scored or conceded. The number of goals scored this season is perhaps too small for useful analysis, though it’d certainly be interesting to see if some teams have rectified certain inefficiencies.
All data from EPL Index, see their Stats Centre for more data.