What is xG in football? Expected goals explained

A 0.20 xG shot is a chance that the model estimates would produce about 20 goals per 100 comparable attempts, on average. It can still be scored or missed this time. The decimal describes a probability; a player cannot score a fifth of a goal.

Four misses do not make the fifth shot certain to go in. Equally, a missed sitter keeps its pre-shot xG, and a long-range goal does not become a 1.00 xG chance afterwards. The chance and its outcome are different pieces of information.

A single shot's value sits between zero and one, but a total can exceed one. Three shots assigned 0.10, 0.20 and 0.40 add up to 0.70 xG. A team recording 2.30 xG has accumulated a total across its shots; it has not earned 2.3 goals or a 230% chance of scoring.

Providers train statistical models on recorded shots and whether they became goals. Common inputs include distance, shooting angle, body part and the action before the attempt, such as a cross or through ball. Some models also describe pressure on the shooter, the goalkeeper's position and the defenders between the ball and goal.

This is why location alone cannot give you an exact xG value. A striker meeting a low pass with an open goal has a different opportunity from a player heading a high cross under pressure, even if both attempts come from a similar spot. The model estimates chance quality from the information it has.

Different providers, training data and model versions can produce different values for the same attempt. Use one provider and a consistent definition when comparing players or teams. The values in the tools here are assigned to teach the arithmetic; moving a slider does not calculate the xG of a real shot.

It depends on what you are comparing. Eight shots worth 0.10 each and one shot worth 0.80 both total 0.80 xG. They have the same expected number of goals in this teaching example. They do not have the same range of possible results: the single shot can produce at most one goal, while eight shots can produce several.

If those eight attempts are independent, meaning one outcome does not affect another chance, their probability of producing at least one goal is about 57%. The one 0.80 xG attempt has an 80% chance. The larger set can reach the same average partly through outcomes with more than one goal.

Keep the two calculations separate. Total xG is the sum of the shot values. The chance of at least one goal is 100% minus the chance of no goals. Multiplying the individual no-goal probabilities requires the independence assumption used in this tool; simply adding xG values does not. Neither calculation says which team will win a real match.

Consider a fictional match: Blue lose 0-1 despite recording 1.80 xG to Gold's 0.60. Blue accumulated the larger total, but Gold scored the only actual goal. Rounding the xG to 2-1 does not reveal a hidden correct score or establish that Blue deserved to win.

Finishing, goalkeeping and ordinary variation can all create a gap between goals and xG. A total also hides how chances were distributed and when they occurred. A team protecting an early lead may play differently and allow more later shots; the full-time xG does not show that sequence by itself.

Use the total as a reason to inspect the attempts, not as a verdict on every player. Look at the chance locations, whether one penalty dominates the total, the shot outcomes and the changing score. A high total says more about the chances recorded than a bare shot count, but it does not measure every part of the performance.

The league table uses actual results. An xG advantage does not add points, change goal difference or break a tie in the standings. Any dated match example on this page uses a separate match record; the two teaching tools use fictional inputs.

Ordinary xG uses the situation before and at the moment of the attempt to assess the chance. Post-shot models add information about how the attempt was executed. Opta's expected goals on target, xGOT, evaluates on-target shots using the underlying chance quality and information including the ball's path and its placement in the goalmouth.

A difficult chance struck towards a corner can have low xG but a higher xGOT value. A strong chance hit tamely towards the keeper can move the other way. Blazing that strong chance wide does not erase its ordinary xG, but it contributes no positive xGOT. xGOT is also not just the pre-shot xG values of the shots that happened to be on target.

Provider definitions matter here too. Stats Perform's updated xGOT adds goalkeeper-position information and details such as deflections, mishits and swerve. Other providers use terms such as post-shot xG or PSxG and their own inputs. Treat those labels as related measures, not interchangeable numbers or one universal formula.

Post-shot information helps assess the difficulty of a goalkeeper's saves. Opponents' pre-shot xG alone cannot do that job: it includes chances that were sent wide or blocked before reaching the keeper. Even a post-shot measure concerns shot-stopping; it does not by itself rate crossing claims, distribution or the whole performance.

Penalties normally count in total xG. Non-penalty xG, often written npxG, removes them, which helps separate a player's penalty opportunities from the rest of their shots. The assigned penalty value is model-specific: Opta's August 2023 explainer gives 0.79, while Hudl Statsbomb's explainer gives 0.78 for its 2022 model update. These are dated model examples, not a universal value or a guarantee for a particular taker.

Ordinary shot-based xG includes more than attempts on target. A recorded shot blocked by a defender or sent wide can still carry a pre-shot value. A dangerous pass that nobody turns into a shot creates no shot xG of its own. So a low xG total can miss attacks that nearly became attempts; other measures are needed to describe that threat.

Rebounds show why adding shot values is different from calculating a possession's chance of ending in a goal. A follow-up may exist only because the first shot was saved or blocked. Those opportunities are connected: if the first goes in, the same rebound is never taken. Do not put their observed values into the independent-chance calculation and present the answer as the probability of that real attack or match.

First, check what the number covers: one shot, a player, a team, one match or a season. Check whether it includes penalties, and whether it is a total or a rate per 90 minutes. An xG per 90 figure adjusts for playing time; it does not say that a substitute will play 90 minutes next time. Keep the provider consistent.

Next, read goals alongside the chances. A fictional player with 3 goals from 1.50 xG is 1.50 goals above their recorded xG for that period. That is a description of what happened, not proof that the finishing rate will last. A player below xG is not owed goals, and a player above it does not have to miss future chances to balance the account.

Then look beyond one match. Repeatedly getting good opportunities is useful information, but a small sample can swing sharply with one penalty, save or exceptional finish. There is no universal xG total that makes a performance good: the competition, opposition, minutes and chances matter.

Finally, keep the stat separate from selection. A player's recent xG does not establish that they will start the next match. Use confirmed team sheets for actual starters, clearly labelled predictions before confirmation, and match records for what happened. The links below let you make those checks without treating this guide's assigned examples as live club statistics.