Lewis Ambrose

What is ‘expected goals’ and why do you need to know?

Expected goals (or ‘xG’) are going to become important this season, so time to brush up on what it means for the game.

What is it?

The statistical measure has been used online and at Premier League clubs for a while now – Arsène Wenger first mentioned it back in October 2015 – so teams can measure how well they’re performing.

For a while we have seen stats flash up at half-time, full-time, or after games listing the number of shots a team had. The problem is, not all shots are the same. You’re much more likely to score from 10 yards than 30 yards.

That’s where expected goals comes in. A statistical measure

Stats can be off-putting but it makes sense: you’re more likely to score the closer you shoot from, you’re less likely to score from tight angles, shots with your feet are

So how is expected goals calculated?

Using over 300,000 shots from the past as a basis, each shot in a game is assessed.

Factors such as type of shot (regular, header, volley), location of the shot, type of shot assist (lay-off, cross, through ball) and how long a team has been in possession are taken into account.

In the end the shot is given a value. You would expect a shot worth 0.2 xG to be scored 20% of the time.

Why should you care?

Because it’s about to go mainstream.

Sky have introduced xG into their work this summer and BBC have announced it will feature on Match of the Day this season.

Clubs are clearly paying attention too, with teams across the league taking fewer and fewer ‘low xG’ shots each season.

While total shots has fallen that is entirely due to fewer from outside the box

via Sky Sports


Of course, this isn’t perfect. The xG models we have today can’t account for the positioning of defenders or goalkeepers, meaning some shots receive strange ratings. Different models from different statisticians don’t quite agree with each other either.

The models also don’t currently manage to account for the ability of the player taking the shot – Lionel Messi is clearly much more likely to score any given shot than Javier Mascherano.

However, expected goals is a better predictor of future performance than actual results, goals scored and conceded, or even shots scored and conceded.

Every week we talk about who deserved to win each game and we bring evidence into the debate. This is just another, more advanced, part of that.

Expected goals is incredibly useful to determine just how good (or bad) the chances teams are creating and conceding are.

It isn’t going anywhere anytime soon.