If you checked the expected goals (xG) from last weekend’s Premier League matches, you might conclude that the statistic is meaningless.
Out of the 10 games played, xG aligned with the actual outcome in only three cases. As an example, Manchester United (2.27) were beaten 0-1 at home by Everton (0.16).
However, xG does not factor in incidents such as Everton going down to 10 men or the fact that they spent much of the match under heavy pressure.
What it can show is that United failed to generate many high-quality chances from their 25 attempts, with Mason Mount (0.44) and Luke Shaw (0.38) producing the most dangerous opportunities.
If xG appears so inaccurate, should we really be placing so much emphasis on it? Perhaps not so quickly.
In week 11, xG aligned with seven match outcomes, and in week 10 it predicted eight correctly.
Since the 2021-22 season, xG has produced an accuracy rate of 59%, and this season it sits at 57.5% — only a one- or two-result difference from what is considered normal.
Crucially, xG is not designed to predict individual match results but to evaluate the quality of chances a team creates and how well they finish them.
Based on xG alone, Aston Villa, Sunderland and Tottenham would all sit in the bottom four; in reality, actual goals place them fourth, seventh and ninth respectively.
Villa have excelled at scoring long-range efforts (more on that later), while for Sunderland and Spurs, goalkeepers Robin Roefs and Guglielmo Vicario have been outstanding in preventing high-quality opposition chances.
So you may wonder: what’s the point of measuring xG at all?
Here’s a clearer way to understand it: if you walk away from a match saying “I can’t believe we lost that” or “if we keep playing like that we’ll be alright”, your team likely posted a higher xG than the number of goals scored.
If your reaction is “I can’t believe we won that”, your team probably had a low xG.
The principle is that, over time, a team’s results should begin to reflect their xG — positively or negatively. This cannot account for factors such as dips in individual form or overall team momentum. For example, Liverpool’s xG this season suggests they should be in fifth place rather than 12th.
