13 January 2020 • 2:59 GMT
Whether we like it or not, stats, data, mathematics are now a big part of football. How do we explain a doctor of particle physics being an integral part of a football club? Someone said, " football isn't rocket science but scientists are driving it in that direction". Some fans see data revolution as a threat rather than as a tool. Unsurprising, I fall into this group probably because I never really liked Maths - I mean, it was a miracle if I had a B in maths and if I had a D, it was expected. But why would some see it as a threat? Simple. Because when simple things become too complex, we displace the simplicity of it. Football is a simple game but where is the simplicity when I have to be a maths guru to understand it?
One of the evangelists of data in football is Matthew Benham, owner of Brentford Football Club who is also the owner of SmartOdds, a betting company. Of no doubt, betting companies make so much use of data, stats and Benham has also brought in the ideology to his club. As much as it hurts to admit, Brentford have been a successful club being run on this data revolution and are closing in on promotion to the Premier League.
There are many data metrics in football today. Some are used to analyse games - xG, xA, PPDA - while some are used in scouting and recruiting players - moneyball. So today's piece of writing is on xG, expected goals.
In football, every shot that a player shoots is given a quality value of 0-1. This value tells us if the shot should have become a goal clearly or if it was to difficult to become a goal, just like probability in maths. Before the invention of Expected Goals, there was an assumption that in every 10 shots a player takes, at least one must result in a goal. But this assumption has become outdated because there is no account of what types of shots they were or how far the shots were from goal. So what then is an expected goal?
An expected goal, xG is a data metric that measures the probability of a shot resulting in a goal using a considerable number of factors. Factors like, the distance of the shot from the goalpost, the angle of the shot( a tight angle or an open angle), was it a 1V1 situation, was it an headed goal or not and so on. There are different xG models today. While some take into consideration whether the player took the shot with his stronger foot or with the weaker, some don't. Some also take into consideration if the player was closed down by several players while taking the shot and some don't. Like I said or wrote, there are different xG models. There is no xG model that takes the player taking the shot into consideration, if it was Messi or Batshauyi. There is no such bias. According to the Opta model, the closer the shot is to the goalpost, the higher the xG.
In this scenario, Tammy Abraham scored this goal. According to the Opta model, the shot had an xG of 0.10 which is a low value xG. A low value xG shot is less likely to result in a goal. The shot had a low value xG because Abraham was closed down by two players and the angle of the shot was a tight one. This means Abraham outperformed his expected goal by 0.90. An xG of 0.10 also means that in every 10 shots, only one would result in a goal. An xG of 0.02 means that in every 100 shots, only two would result in a goal. An xG of 0.80 means in every 10 shots, 8 would result in a goal. Simple, isn't it?
This season after 16 appearances, Heung Min Son has racked in 12 premier league goals but his xG of 5.39 according to the Understat model tell us that he should have scored 5 goals and not 12. This means Son is outperforming his xG by 6.61 (12 - 5.39). In simple words, Son should have only scored 5 goals but he has scored 12. This makes Son a very clinical striker and probably the most clinical in the Premier League. But let's not get carried away, this could mean Son is a very clinical striker or that Son has just been lucky. Also, it is quite impossible for a player to keep outperforming his xG by this margin.
Also according to the Understat model - I use the Understat model a lot, well because it's free - Sheffield United have scored just 8 goals this season despite having an xG of 16.8. This tells us they are underperforming their xG by 8 goals. In simple words, they should have scored 16 goals but they have only managed to net 8. This means that Sheffield have really poor strikers or that their strikers have been extremely unlucky.
There is more to xG as we have xGA (expected goals against), NPxG(Non Penalty expected goals) and so on but in order not to make this look like a maths tutorial, I have decided to round it off here. Don't fret if you didn't quite get today's article, you might just read it all over again or just join the "get rid of stats" campaign.
No comments:
Post a Comment