What is xG?

28.12.2020

In the past decade, the way Football was being analyzed changed drastically. Coaches, pundits and journalists started to look at the statistical side of the beautiful game, something that we have only seen in American Sports thus far. One particular stat made it into the mainstream like no other - expected goals or xG.

xG is a measurement that represents the probability of a shot resulting in a goal, going from 0 - which would mean that scoring a goal is impossible - to 1 - which is a guaranteed goal. It measures the quality of a goal-scoring opportunity, for example, if a chance has an xG of 0.9, you would expect the player to score that chance 9 out of 10 times. What makes expected goals so interesting is that it´s a rather non-conventional stat. It can´t be measured like the number of shots or amount of yellow cards, because, it doesn´t really exist. xG is calculated by stats companies, who use large data-sets and machine learning algorithms to determine how likely the players is to score.

Whilst this sound very complicated, xG is just giving us information, that we already know. When football fans say "they should have won this game" "we were the better team" or "he should have scored that" what are they basing those statements on? Well, if you watch a lot of football you know from experience that if a team creates way more chances than their opponents, they win more times that they don´t. You know that because you watched plenty of games where that was the case. That is exactly how xG is calculated. The number then tells you how many goals a team should have scored, based on their chances. On a similar note, when your star-striker just missed an absolute sitter, you are going to be upset and rightly so, because you have probably seen him score that same chance loads of times in the past. But not all shots are the same, xG takes into account, where the player is located, how far away from the goal he is, at which angle he is shooting and the number of defenders, who could potentially block the shot. After all, tap-ins are easier to score than 30 yard-screamers.

However, expected goals aren´t real goals. When a team clearly created more high-quality chances but failed to score one of them, while the other team scores a late goal, despite the fact that they only sat back the entire game, they could have way more expected goals, but still lose the game. When Liverpool beat Crystal Palace 7-0, the xG-score was 2.87 to 0.49, which suggests that Liverpool should have won, but not by that many goals. Another example would be Manchester United, who drew 2-2 with Leicester but should have beaten the Foxes by one goal, based on xG. xA (expected assists) and xP (expected points) work in a very similar way but are not referenced as often.

author: Dominik Plöchl