xG Against: How to Read Expected Goals Conceded

xG against, usually written xGA, means expected goals against: the total expected-goals value of the shots a team allows its opponents to take. It measures the quantity and quality of chances conceded, rather than the goals that actually go in. The Premier League defines xGA as a measure of how likely conceded chances were to become goals.

Lower xGA is generally better. A clean sheet with low xGA suggests the team restricted its opponents’ chances. A clean sheet with high xGA means those chances existed but did not become goals—a different defensive story.

How is xG against calculated?

An expected-goals model assigns each shot a probability of becoming a goal, from zero to one. A 0.20 xG chance has an estimated 20% scoring probability. Models use historical shots and characteristics such as distance, angle, body part and the preceding pass. Depending on the provider, they also account for defender pressure and goalkeeper position. StatsBomb explains these inputs and why models differ.

For the defending team, those shot values are added together:

xGA = sum of the opponents’ shot xG values

In a simplified example, allowing three shots worth 0.10, 0.20 and 0.40 xG produces 0.70 xGA. That total is unchanged whether the opponent scores zero, one or two goals from them. Within the same model and match coverage, one team’s xG is the other team’s xGA.

This is a shot-based total, not just a measure of shots on target: a missed chance still contributes its pre-shot scoring probability. Nor does 0.70 xGA mean exactly 0.70 goals “should” have been conceded. As Opta explains, expected goals measures chance quality, not a guaranteed result.

What xGA tells you that goals conceded cannot

The distinction is visible in the Premier League’s published 2022/23 figures:

Team	Goals conceded	xGA	Goals conceded minus xGA
Manchester City	33	33.07	−0.07
Newcastle United	33	40.09	−7.09
Fulham	53	65.56	−12.56

Source: Premier League’s 2022/23 xGA review.

City and Newcastle conceded the same number of goals, but City allowed less expected scoring value. Their identical goals-against records therefore conceal different levels of chance prevention.

Fulham conceded 12.56 fewer goals than their xGA. That is a favourable gap between chances allowed and actual outcomes—not evidence that they prevented chances better than City or Newcastle. To judge chance prevention, compare xGA itself; to judge the scoreboard, compare goals conceded.

xGA is a team measure, not a centre-back rating. Pressing, midfield protection, possession retention and goalkeeper positioning can all affect the shots a team allows.

Check the denominator and the match situation

A season’s xGA is a total, with no denominator. For teams with different numbers of matches, use xGA per match: total xGA divided by matches played. If a table reports xGA per 90, its denominator is playing time expressed in 90-minute units. Check the provider’s playing-time convention and whether extra time is included rather than treating all totals as equivalent.

Then examine how the chances arose. In Arsenal’s 2–2 draw at Manchester City in 2024/25, Arsenal played the second half with ten men. Arsenal’s own analysis reported that only five of City’s 33 shots exceeded 0.10 xG. That illustrates why shot count needs chance-quality context—and why a defensive performance under a red card should not be treated as a normal eleven-versus-eleven match.

For comparisons, keep the provider and competition consistent, check whether penalties are included, and prefer a run of matches to one result. Non-penalty xGA excludes penalty shots; it still includes non-penalty set pieces, so it is not an open-play-only measure. Use it to separate the chances allowed outside penalties, without ignoring the penalties’ effect on actual results.

xGA is not a goalkeeper rating

Conceding fewer goals than xGA can reflect saves, poor opposition finishing and variation in outcomes. The gap alone cannot tell you how much credit belongs to the goalkeeper.

For shot stopping, post-shot expected goals is more appropriate: it accounts for the execution of shots on target, including where they travel in the goalmouth. Stats Perform distinguishes this from pre-shot xG. Our guide to goals saved above expected explains that separate comparison.

Finally, standard shot-based xGA does not value an attack that ends without a shot. A dangerous pass that nobody reaches leaves no shot value to add. Read xGA alongside the shot map and match context: it answers “How much scoring opportunity did this team allow through shots?”, not every question about how well it defended.