It’s no surprise that Mesut Özil, Arsenal’s mercurial playmaker, has been grabbing the headlines. He has 10 assists after featuring in just 11 matches, and he’s the new owner of the Premier League’s record for consecutive matches with an assist. But a closer look at his statistics shows that he may be starting to go downhill – albeit from a very high peak.
If Özil’s only job were to assist goals, then there’d be no point writing this post. His real job, though, is to make Arsenal better than anyone else would at the same position on the field. That means he has to support the collective effort, helping other players to make assists as well – not to mention defending. If he finished the season with a mammoth total of 30 assists, but no one else had even one, then Arsenal would probably be relegated. So how is he fitting into the team as a whole?
One of NYA’s tools for answering this question is the Shapley value, which measures a player’s overall contribution to winning. It captures a player’s intangible benefits or disadvantages as well as anything that can easily be counted on the field. In both 2013-14 and 2014-15, Özil’s Shapley value stayed pretty close to the average across Arsenal’s squad (weighted by minutes played), with the equivalent of fewer than 25 full matches played in each season. In other words, he was no more or less pivotal than other players at the club, even as the club’s level of success varied. And this situation appears to have persisted.
To see this with numbers, we can transform the Shapley value into a sort of X-factor to see whether a player is outperforming his teammates. For example, if Shapley values are on a scale from 0 to 100 across the league, we can use the following formula:
player’s X-factor = [0.5 * (player’s Shapley value – team’s average Shapley value)] + 50
This formula yields X-factors that are also on a scale of 0 to 100 (though the actual range is slightly smaller, since the team average includes the player in question). A player with a Shapley value of 65 on a team whose average is also 65 has an X-factor of 50 – just at par, since he’s not doing anything exceptional relative to his teammates. But a player with a Shapley value of 90 on a team where the average is 20 has an X-factor of 85; clearly he’s making a tremendous difference when he comes onto the pitch.
Özil’s X-factor through 12 games of the 2015-16 season was 49. Last season, it was just over 52. This isn’t a big difference, but it suggests that Özil isn’t supplying anything special to the Gunners that he wasn’t already bringing last term.
That’s not the end of the story, though. NYA typically uses half a dozen metrics to assess different aspects of a player’s game, adjusting for factors including the quality of his teammates. Two of them are listed below. Özil’s contribution to creating shots – and preventing the opposition’s shots – has declined, even as Arsenal’s overall rating has risen. And his ability to advance the ball into dangerous areas – and to stop opponents from doing the same – has slipped by even more. He used to be one of the top few players in the league, if not the world, by this last metric. He may not be anymore:
Even the volume of Özil’s attacking touches has gone down by about five per 90 minutes. Indeed, the fact that his Shapley value has stayed as high as the Arsenal average, despite his shot creation and ball progression numbers dropping, suggests that he has found other ways to help the team. Watching video and interviewing Özil’s teammates and coaches might reveal more about how his contributions have changed.
We should interpret these data with caution, since the 2015-16 season is still young; not all teams have played each other, so strength of schedule is likely to be an issue. Regardless, Özil continues to be one of the best players in the Premier League – we just shouldn’t read too much into his assists. In truth, he’s 27 years old and may just be starting to shift into a different phase of his playing career. That’s something for Arsenal to think about as the club plans for the future.