Introducing Advanced Statistics Part 1: Possession

This will be the first in a series of posts in which I attempt to explain some of the statistics I've created and why I think they are more useful than the traditional statistics.

One of the things I've tried to do with my statistics is separate out the different skills needed in soccer. As I see it, there are three skills needed on the offensive side of the ball: possession, creating chances to score, and turning those chances into goals. Similarly, on the defensive side of the ball the three skills are limiting opponents' possession, limiting opponents' scoring chances, and limiting the taking of those chances. The statistics I've created evaluate a team's skill in each of these areas. In this post, we'll focus on those evaluating possession. 

Possession is easy to calculate as it is provided in nearly every match report, with pundits often pointing to possession statistics as evidence of how a team played. In reality, there is no correlation whatsoever between possession and winning, but that doesn't mean possession is useless. Looking at possession can tell you how good a team is at keeping the ball and winning it back from their opponents.

Ideally, I'd like to look at the number of times a team had possession too, similar to what Hollinger does for basketball. Time per possession would tell us how good a team was at keeping the ball; time per opponents' possession would tell us how good a team was at winning the ball back. I think that would be much more useful analytically. However, the best measure of possession available to us is the Time of Possession statistic, or the percentage of time a team had the ball in the match. Since possession is zero-sum, we don't really need to calculate Opponents' Possession; in other words, we know from a team's possession how successful they were at limiting their opponents' possession. Last season, the league mean for possession was 50% (duh) and the range was from 60% (Arsenal) to 39% (Stoke).

What good is it to look at a team's possession if it is not correlated with winning? Well, for one thing it can tell us something about a team's approach. Take for example Stoke's game against Chelsea last weekend. Stoke had just 34% of possession. That tells us that Stoke was sitting back and absorbing pressure and that Chelsea was initiating most of the play. What it does NOT do, and what many commentators mistakenly think it does, is indicate that Chelsea was the better side. As stated above, last year Stoke averaged 39% possession and yet finished in 12th place. Possession is just one piece of the puzzle, the rest of which we will get to later.