How xG Works in Football: A Clear Guide

How xG works, step by step

Record a shot. An xG calculation begins when a player attempts a shot. The provider defines which events count as shots and how it treats unusual cases, so definitions can vary.
Describe the situation. Data about the attempt may include where it was taken, its angle to goal, the body part used and whether it followed a pass, cross, rebound or set piece.
Compare with historical examples. A statistical model uses past shots with similar recorded circumstances to estimate the chance that the attempt will be scored.
Assign a probability. The model gives the attempt an xG value from 0 to 1. A value of 0.20 means an estimated 20% chance of a goal for a comparable attempt, not a guarantee about this shot.
Add the values. A team’s xG for a match is generally the sum of its shots’ xG values. Comparing that total with goals scored helps describe chance creation and finishing, but does not explain every part of a performance.

What xG measures—and what it does not

xG stands for expected goals. It is a probability estimate for shots, built from patterns in a collection of previous attempts. The central question is: given the circumstances recorded for this shot, how often did comparable shots result in goals?

Suppose a team takes three shots with values of 0.50, 0.25 and 0.10. Its total is 0.85 xG. That does not mean it should score exactly 0.85 goals, or that the team is guaranteed to score. It means the model assigns those attempts a combined expectation of 0.85 goals. In one match, the team might score none; in another, it might score two.

xG estimates the quality of chances, not the quality of every action in a match. It does not directly measure possession, passing accuracy, defensive pressure, a player’s overall ability or whether a team deserved to win. Those questions require other evidence and context.

Which details affect a shot’s xG?

The exact inputs depend on the provider’s data and model. Common factors include:

Distance and angle: Shots nearer the goal and with a clearer view generally have higher probabilities than distant attempts from a tight angle.
Shot type: A header and a shot with the foot may have different outcomes in comparable situations.
How the chance began: A through ball, cross, rebound, corner or other set piece can provide useful context if the model records it.
Defensive pressure and players’ positions: Some models account for defenders, the goalkeeper’s position or the amount of pressure; others may not have these details.
Other event details: The model might consider whether the attempt followed a particular kind of pass or came during a one-on-one situation.

These are not universal ingredients. Two websites can assign different values to the same shot because their models use different data, definitions and methods. A model based mainly on shot location may not distinguish between a clear attempt and one blocked by a defender as precisely as a model with richer tracking information.

What to check when reading an xG figure

Before comparing numbers, check that they refer to the same match, competition and type of xG. In particular, look for whether the figure includes penalties and how the provider defines the statistic. A penalty is usually assigned a relatively high, fixed or near-fixed value, but its exact value varies by model.

Also check whether the number is for a team, an individual player or a single shot. Team xG is usually a sum of shot values. Player xG usually totals the value of that player’s attempts. Neither figure is necessarily adjusted for playing time unless it is explicitly presented as a rate.

For a useful comparison, consider more than the final total:

Look at the number and locations of shots, not only the total xG.
Separate penalties from open-play chances when the data allows it.
Compare the same provider’s figures where possible, because models are not interchangeable.
Use match context: a team protecting a lead may take fewer risks and create fewer chances.

Common misunderstandings and limitations

“A 0.80 xG chance should be a goal.” No. It represents an estimated 80% scoring probability for a comparable shot. The attempt still has a substantial chance of not going in, and the value is not a prediction of certainty.

“The team with higher xG must have played better.” Higher xG often indicates that a team created more or better shot opportunities, but it is not a complete verdict. It can leave out build-up play, defensive work, game state and details the model cannot observe.

“xG proves whether a player is a good finisher.” A gap between goals and xG can describe what happened over a period, but a short run may be strongly affected by chance. Assessing finishing ability requires more attempts and care; xG alone cannot establish why a player scored above or below the estimate.

There are also limits in the underlying data. If a provider does not record a factor such as pressure or goalkeeper position, its model cannot account for that information directly. Models can be calibrated differently, and their estimates should be treated as useful summaries rather than exact measurements of reality.

Goals are the actual outcome; xG is an estimate of the chances behind attempts. Goals decide the score, while xG can help describe whether the shots were generally promising.

Shots count attempts but treat a speculative effort and a close-range opportunity as equal. xG gives attempts different weights according to their estimated scoring probability.

Expected goals on target (xGOT), where offered, assesses shots that are on target and may include information about where the ball is headed. It is not the same as ordinary xG, which is generally assigned to the shot based on circumstances before its outcome is known. Providers’ definitions still matter.

Frequently Asked Questions

What does 1.0 xG mean?

For a team, 1.0 xG is the sum of its shots’ estimated probabilities, not a promise that it will score one goal. For an individual shot, a value of 1.0 would mean the model treats a goal as certain, though ordinary shots do not normally receive that value.

Can a team win with lower xG?

Yes. xG describes estimated chance quality, not the result. A team can score from a few attempts while its opponent misses better chances, so the team with lower xG can still win.

Why do different sites show different xG?

Providers may use different shot data, inputs, event definitions and statistical models. Their totals can therefore differ even when they describe the same match. Compare figures from one provider rather than assuming all xG values are identical.

Does xG include penalties?

Often it does, but check the provider’s definition. Penalties may be included in a team’s total or displayed separately, and the assigned value can differ between models.

Is xG a prediction of the final score?

No. It summarizes the estimated scoring probabilities of attempts that have already happened. It can inform analysis, but it does not forecast the exact score or guarantee what will happen next.

Sources: FIFA. Menus and buttons can differ slightly by version. Spotted something out of date? Tell us.