Alex Cooke looks at how Swindon’s Bayern Munich-like dominance of the ball predicts big things

This season Swindon have dominated the opposition and the ball. Town are top of the League One and top of the possession table. After 26 games their average possession has been a whopping 58.23%, almost five percentage points more than the Franchise (53.57%) Amazing isn’t it?

Well, yes, and no. The number looks amazing and hugely seductive, but it doesn’t really mean anything. The figure is right, but it is just a figure. We have to be careful about what it actually tells us, and what we use it for. It is a trap I’ve fallen into many times myself…

That number of 58.23% comes courtesy of a fine piece of research and writing by The Hefty Toe. Give it a read, it is well worth it. Just bear a few things in mind while you enjoy…

What can possession statistics tell us?

Very little. Possession figures don’t tell us where the ball was played or how it was played. Hoof, pass, header, mis-control or scuff, all are counted the same. In fact, high possession stats like these can come from games in which the ball is passed relentlessly across the back line, ala Paul Hart, or pinged around inside the opposition’s 18-yard-box. The number doesn’t tell us which is which. Want an example? Remember Paul Hart’s 0-0 draw with Walsall? Swindon had 55% of the ball. Yet it was utterly grim.

The information does exist at the higher levels of the game when ‘possession in the final third’ or ‘completed passes in the final third’ stats give a much better picture and highlight whenever what Jose Mourhino call ‘resting on the ball’ and Arsene Wenger named as ‘sterile domination’.

Does good possession mean better results?

Possession stats also don’t predict if a team will win or lose over a season. Bad teams can dominate the ball and good teams can use very little possession to win games and football is too low a scoring game for possession to provide accurate predictions. Yes, Barcelona won almost everything keeping the ball, but Inter actually won a European Cup deliberating giving it away. However, higher possession will generally mean a team has more shots. And, generally speaking more shots does lead to more goals, but not always.

What this table actually tells us is that the better teams are better at keeping the ball, not that those who keep the ball are automatically better teams. In League One at the moment, three of the top teams who are winning have an ethos of keeping the ball, there are also quite a few who lose a lot who keep the ball too, for example Walsall and Yeovil. Possession simply isn’t a predictor of a good team, it tends to be the by-product of a good team.

What can possession stats tell you about an individual game?

Again, not that much. The example so often pulled out is the Champions League group game between Celtic v Barcelona in 2012. Opta’s own figures show Celtic had the ball just 16.36% of the game, but won 2-1. Bayern Munich similarly dominated Real Madrid last season in the Champions League with 71% over two legs but lost the tie 0-5.

That is an extreme example but it also points to a quirk of possession statistics: teams who are ahead in the game tend to have less possession. Some call it ‘shelling’ – the phenomenon of a team bagging a goal in and dropping deeper in the field and concentrating more on keeping their shape than the ball. Town used to do it under Paolo Di Canio.

How is possession measured?

There are a couple of different approaches to calculating possession with the big data companies. Some use a simple chess clock, stopping and starting it as possession changes between one side and the next. Others, such as Opta, use a more complex calculation based on the number of passes made during the game. The final version, and this one is key for clubs such as Swindon, is based on the subjective judgement of the person logging the key events for the stats. Some might call it an educated guess, some might be less kind – but it does seem to explain why the figure can vary so much. It is understandable though as the alternatives both demand more resources for very little gain.

Can anything predict results then?

Yes, the amount of shots a team has tells us a lot more about how a side will perform over a season. Total Shot Ratio is actually a fairly solid predictive model of a team’s quality, but not a very logical one. There are others, but that can wait for another day…

Image modified from ThisIsSTFC.co.uk