In November 2017, Arsene Wenger talked about a game against Manchester City in terms of expected goals: “If you look at the expected goals, it was 0.7 for them and 0.6 for us, it was a very tight game”. Arsene was later criticized by football analytics community for not using the stats correctly, but general audience wondered what he was talking about.
What the hell is Expected Goals (xG)?
Expected goals is one of the few accepted concepts in football analytics. Simply put, expected goals is a metric that quantifies quality of scoring chances. When we look at a match statistics, we usually can see the score, shots and shots on target. But not all of the shots are created equal. Some shots are better than others. For instance, a shot outside of the penalty box has a much lower probability than a shot from a penalty spot. Expected goals measures the quality of the chances by assigning certain value to each chance based on historical probabilities.
How Do you Calculate Expected Goals?
Lets use a penalty shot as an example. It’s a fantastic scoring opportunity. But its not converted into a goal 100% of time, is it? So how often does it become a goal? If we look at Premier league seasons 2011/2012 to 2015/2016 we will find that there were 443 penalties and only 347 of them turned into a goal. That’s an average of 78.3% conversion. We therefore assign to a penalty an expected goal value of 0.783.
On the other end of a spectrum is a a kick outside of the penalty box. Next time you are screaming “Shoot! Soot!” at a player who comes closer to the penalty line, remember: player’s chance of scoring is only 3%. So every shot outside of the penalty area is only worth 0.03 expected goals.
With all the data available online, you can easily research probabilities of different shot locations. Next, you tally up all of the chances and their values for a given match and voila! You got the total expected goals. You now know how many goals your team should have scored. What makes this concept so attractive is an objective way of evaluating the chances and removing luck out of equation.
This is a very simplistic model of calculating expected goals. Some of the more complex models take into consideration not only the location of a shot, but also the angle of a shot, how many defenders where in front of a shooter, was the shot made with a foot (more likely to score) or head (less likely to score), etc.
Why does a football club need to spend time on tracking expected goals?
Simply put its a non subjective way of assessing performance. It helps make better quality decisions that are fact based. It can help your club answer
- How good are the chances we are creating?
- How successful is our defence at stopping shots from dangerous areas?
- How many goals are we likely to score this season if we continue with the same strategy?
- Train your players to shoot from better positions and avoid low probability shots
- Predict the standing in the table with pretty good accuracy
- Is our team effective in converting goal scoring chances? (what’s the difference between expected goals vs actual goals)
- Rank your players based on the quality of chances they create
- Scout opposition and assess their effectiveness at creating chances
- Should we fire the coach?
Okay, maybe number 9 is a bit of a stretch, but the model has many applications in the club.
Are you using a similar model right now? I would love to learn more about it in the comments below