xG vs xGA Explained: Attacking and Defensive Football Statistics

What is the difference between xG and xGA?

A team can win 2–0 and still leave its coach worried about the chances it conceded. The scoreboard records the goals; xG and xGA help explain the chances behind them. Expected goals (xG) sums the estimated scoring probabilities of a team’s shots, while expected goals against (xGA) sums the corresponding probabilities for its opponents’ shots. Together, these statistics describe the quality of chances created and conceded, rather than just the final score.

If a team records 1.8 xG and 0.9 xGA in a match, it created chances with a combined estimated value of 1.8 goals and allowed chances worth 0.9. These figures are illustrative, not results from a real SmartMatchData match. Neither number is a guaranteed goal total.

How is xG calculated?

A shot-based xG model estimates the likelihood of a shot becoming a goal based on historical examples. Depending on the provider, inputs can include shot distance, angle, body part, assist type, set-piece situation and defensive context. The values for a team's attempts are added to produce its xG total.

A single shot valued at 0.25 xG represents an estimated one-in-four chance of scoring under that model's assumptions. For a broader introduction, see our Expected Goals (xG) guide.

How is xGA calculated?

xGA uses the same general idea from the defending team's perspective. It adds the xG assigned to opponents' shots. When two teams play each other and the same shot model and match scope are used, one team's xG corresponds to the other team's xGA for that match.

A team with low xGA is usually restricting the volume or quality of the shots its opponents manage to take. That is valuable defensive evidence, but it is not the whole story. However, xGA is not a complete measure of defensive ability. Opponent strength, game state, pressing, goalkeeper actions and whether shots are prevented before they occur also affect how a defence should be assessed.

How to interpret xGA as a defensive statistic

Expected goals against answers a specific question: **how much scoring chance quality did the opposition generate?** It does not directly tell us how many goals a goalkeeper should have saved, how often defenders won their duels or how well a team pressed before a shot could happen.

Imagine two hypothetical defences over ten matches. **Team A** concedes **9.0 xGA** and **12 actual goals**. **Team B** concedes **15.0 xGA** but only **8 actual goals**. Team A allowed less estimated chance quality, even though it conceded more goals on the scoreboard. Team B's outcomes were better, but the underlying chances it allowed were more dangerous in aggregate.

Neither comparison proves that Team A has better defenders or that Team B's goalkeeper was solely responsible for the difference. The teams may have faced different opponents, and xG models describe the chances that were taken rather than every defensive action leading up to them.

Low xGA versus a high clean sheet count

A clean sheet means a team conceded zero goals in a particular fixture. Low xGA means the opponents' attempts carried relatively little estimated scoring probability. These observations often align, but not always.

A side may keep a clean sheet despite allowing **2.1 xGA**, perhaps because the opposition missed good chances or the goalkeeper made important saves. Another may concede one goal while allowing just **0.4 xGA**, with a single difficult finish accounting for much of the result.

For defensive analysis, examine **xGA per match**, actual goals conceded and the number of fixtures behind each average. Do not infer the probability of a clean sheet in the next match directly from historical xGA. That requires a projected opposition scoring rate and a forecasting model. Read Clean Sheet Probability Explained for the distinction.

Does goals conceded minus xGA measure goalkeeper quality?

Not on its own. A team's goals conceded minus xGA compares actual goals allowed with the estimated scoring probabilities of shots conceded. It can identify an interesting gap, but the gap combines several influences: finishing by opponents, goalkeeper actions, shot placement, model assumptions and ordinary variation.

Standard **pre-shot xG** is based on the situation before the shot's final execution. **Post-shot expected goals**, where available, can incorporate the placement and trajectory of shots that reach the goal area and may be more directly useful for evaluating shot-stopping. The two metrics are not interchangeable, and post-shot data is not necessarily available for every SmartMatchData competition.

A goalkeeper's performance should not be judged solely from goals conceded minus team xGA. Similarly, a defence that repeatedly restricts opponents to low-quality attempts deserves recognition even when an occasional difficult shot becomes a goal.

Defensive xGA per match: an example with unequal schedules

Suppose Club North allows **12.0 xGA in eight games**, while Club South allows **18.0 xGA in fifteen games**. North has the lower total, but its average is **1.50 xGA per match**. South's average is **1.20**.

South has allowed less estimated chance quality per game in this hypothetical comparison. However, it has played more fixtures, and the opponents may differ. A fair conclusion reports both the per-match rate and the sample size rather than ranking defences by season totals alone.

For a fuller team-comparison process, see How to Compare Football Teams Using Statistics.

What rising or falling xGA can reveal

An improving defence may allow fewer shots, force attempts from worse positions or prevent opponents from creating clear opportunities. Those changes can appear as falling xGA, even before clean sheets become more frequent.

But one match with a penalty, a red card or an unusually strong opponent can change the average sharply. Check the chronological sequence, compare recent fixtures with the season baseline and inspect whether the trend survives differences in venue and opposition.

For instance, an illustrative five-game xGA sequence of **1.8, 1.5, 1.2, 0.9 and 0.7** suggests improving chance prevention. It does not establish a permanent tactical transformation without more context. The attacking and defensive form guide shows how to evaluate those changes over rolling windows.

A practical xGA checklist for match analysis

Start by confirming that the match is complete and the xG data is actually available. Identify the opponent and venue. Compare the team's xGA with its goals conceded, but keep shot prevention separate from goalkeeping outcomes.

Next, look across several fixtures rather than one result. Check whether a single penalty or unusual match drives the average, whether the opponent was particularly strong and whether home-away splits are based on enough games. Finally, compare xGA with attacking xG so that a defensive improvement is not assessed in isolation from changes in the team's overall chance balance.

When browsing Statistics or Leagues, treat unavailable xGA as missing data, not zero. A useful conclusion states the period, sample size and actual metric: *the team allowed fewer expected goals per match over this set of fixtures*. It should not claim a guaranteed clean sheet or a precise next-match probability.

What is xG difference?

xG difference is xG minus xGA. A positive value means a team generated more estimated chance quality than it allowed during the measured period. For example, 1.8 xG minus 0.9 xGA gives an xG difference of plus 0.9. A negative value means the reverse.

This statistic is useful for summarising a balance between chance creation and prevention. It does not determine who deserved to win an individual match, nor does a positive xG difference guarantee positive results in the future.

Per-match averages versus season totals

Cumulative xG and xGA increase as teams play more matches. Comparing season totals for teams with different numbers of completed fixtures can therefore be misleading. Per-match figures divide each total by the number of qualifying matches in the sample.

Suppose one team has 18 xG in ten games and another has 20 xG in fifteen. Their xG per match values are 1.8 and approximately 1.33 respectively. This is a hypothetical example. Make sure the underlying matches, competitions and provider definitions are comparable before drawing conclusions.

Home and away splits

A team's chance creation and chance prevention may vary by venue. Comparing home xG and xGA with away xG and xGA can add context to an upcoming fixture. But venue splits reduce the sample size and can amplify the effect of a few unusual matches.

Consider the strength of opposition and whether a team has played a balanced schedule. The Football Form Statistics guide explains how to interpret recent results and home-away records alongside underlying metrics.

xG, xGA and actual goals

Goals for and goals against are observed outcomes; xG and xGA are model estimates of chance quality. A team can concede few actual goals while allowing relatively high xGA because of goalkeeping, missed chances or variation. Conversely, a team may score more goals than its xG over a short stretch.

Such differences can be informative but should not automatically be labelled luck or proof of sustainable finishing skill. A larger sample, tactical context and the quality of the shot data are important. Different providers may assign different values to the same chances.

Can xG and xGA predict the next match?

Historical xG and xGA can be useful model inputs, but they are not themselves pre-match probabilities. A forecast must also account for opponent strength, home advantage, sample recency and other factors. It must not use information that became available only after kickoff.

SmartMatchData separates observed football statistics from its locked pre-match model probabilities. Read the Model Methodology and Poisson and Dixon-Coles guide to understand how scoring-rate models can translate information into outcome probabilities.

Check the data before comparing teams

Not all leagues, seasons and matches have equally complete xG coverage. Missing xG values must not be interpreted as zero. Check whether figures include penalties, whether the same provider and model version are used, and how postponed, abandoned or incomplete matches are treated.

Use the Statistics page and Leagues directory to explore available data, while keeping sample sizes and coverage in mind. xG and xGA are useful analytical evidence, not guarantees of results or financial returns.