SportQuantix Guide · Soccer
Expected Goals (xG) Explained
Expected goals, or xG, is a soccer statistic that estimates the probability of a shot becoming a goal. A model considers the shot's recorded characteristics and patterns in historical data to estimate how likely it is to score. In football, xG helps explain chance quality rather than simply counting attempts. It does not guarantee a goal or tell you who will win. [1][2]
That distinction is useful whenever the scoreboard and the chances seem to tell different stories. A team can create several promising opportunities and fail to score, while its opponent converts a difficult shot. xG offers another way to understand what happened, without replacing the actual result.
Key takeaways
- xG estimates a shot's scoring probability, not a guaranteed goal.
- Adding shot values gives an expected goal total, not the probability of scoring or winning.
- There is no universal xG formula. Compare figures from the same provider and check the sample.
- Post-shot measures add information about the attempt after contact and answer a different question.
What does an xG value mean?
A shot's xG value sits on a probability scale from 0 to 1. You can express the same estimate as a percentage. [1][2]
For example, a hypothetical shot valued at 0.20 xG has an estimated 20% probability of becoming a goal under the model, given the characteristics it records.
That does not mean this particular shot will score. Nor does it guarantee that exactly one of the next five similar shots will go in. Probabilities describe uncertainty. Even attempts that receive the same estimated value can have different outcomes.
Think of xG as an estimate of the opportunity, not a goal that a player or team is owed.
How is xG calculated?
An xG model learns from historical shots and their outcomes. It uses recorded information about a new attempt to estimate the likelihood of scoring. There is no universal xG calculation formula that every provider uses. [1][2]
Adding a team's shot values is straightforward arithmetic. Producing the value for each shot is a separate modeling task. You cannot reliably turn a pitch location into an xG value with a simple rule about distance alone.
Common information used by xG models
Hudl's explanation identifies several inputs commonly used across models: [1]
- Distance to goal: where the attempt is taken from.
- Angle to goal: how much of the goal is available from that position.
- Body part: whether the attempt is a header or a shot with the foot, for example.
- Preceding action: whether the chance follows a cross, through ball, set piece or dribble.
These details help distinguish attempts that a raw shot count treats equally.
Provider-specific context
Some models include more detailed information about what surrounds the shooter. Hudl describes StatsBomb features including goalkeeper position, the positions of players in the shot's frame and the height at which the ball is struck. [1]
Opta's methodological explanation also lists goalkeeper position, defensive pressure, other players' positions, shot type and pattern of play among its inputs. [2]
These are examples of documented provider methods, not a checklist that applies to every xG figure. Different data collection and modeling choices can produce different estimates for the same attempt.
How individual shots add to team xG
A team or player's xG total is the sum of the relevant individual shot values. [1][2]
Hypothetical three-shot example
The following values are assigned for illustration only. They are not actual model calculations or match statistics.
| Shot | Illustrative xG value | Equivalent estimated scoring probability |
|---|---|---|
| A | 0.05 | 5% |
| B | 0.20 | 20% |
| C | 0.35 | 35% |
| Total | 0.60 xG | Expected total, not a scoring probability |
Illustrative locations and assigned values, not real match statistics or model-generated outputs.
- Shot A
- 0.05 xG
- 5% probability
- Shot B
- 0.20 xG
- 20% probability
- Shot C
- 0.35 xG
- 35% probability
Hypothetical expected total
0.05 + 0.20 + 0.35 =0.60 xG
The total is an expected number of goals, not a 60% chance of scoring or winning. These pitch locations do not produce the assigned values.
Total xG = 0.05 + 0.20 + 0.35 = 0.60
The total describes the expected number of goals associated with those three estimated shot probabilities. It does not mean the team scored 0.60 goals. Actual goals are whole numbers.
It also does not mean there is a 60% probability of scoring at least once, or a 60% chance of winning. Those are different questions. Summing expected contributions does not require assuming the shots are independent, but calculating the probability of at least one goal requires additional assumptions about how the attempts relate to one another.
The shot diagram's locations are illustrative too. They do not imply that a shot from a particular location always receives that value.
xG versus actual goals
Actual goals record what happened. xG estimates the quality of the shooting opportunities using the model's available information.
A goal can come from an attempt with a low estimated probability. A high-quality chance can be missed or saved. That is not a contradiction: neither a low probability nor a high probability makes an outcome certain. [2]
Why can a team have higher xG but score fewer goals?
Consider this separate hypothetical match, not a real result:
| Team | Illustrative xG total | Actual goals in the hypothetical example |
|---|---|---|
| Team A | 1.60 | 0 |
| Team B | 0.70 | 1 |
Team B wins in this example. Team A's higher xG indicates a larger estimated scoring expectation from its recorded shots, not a right to the victory.
Finishing, saves, the details a model does not capture and ordinary variation can all contribute to a difference between xG and goals. The aggregate totals alone cannot isolate the cause. Opta explicitly cautions that higher match xG does not establish that a team should have won. [2]
There is no requirement that a team which scored more than its xG must score less later to make the totals balance. Past outcomes are already recorded; they do not create a debt that future matches must repay. [2]
How to interpret team xG
Team xG can add context to the chances a side creates. The opponent's xG can similarly describe the estimated quality of chances it allows. Neither figure is a complete account of team performance. [1][2]
Useful questions include:
- Did the team generate many attempts, or a smaller number of higher-value chances?
- Was the total concentrated in one opportunity or spread across several shots?
- Are the figures from the same provider and using comparable coverage?
- Are penalties included, or are these non-penalty totals?
Read the numbers alongside the match. A side chasing a goal may play differently from one protecting a lead. An early red card, the opponents faced and the balance of open-play versus set-piece chances are relevant context. They are not necessarily adjustments already built into the displayed total.
For broader background, read Sports Analytics: Understanding the Game Through Data.
How to interpret player xG
A player's xG total summarizes the estimated quality of the shots credited to that player. It can help distinguish shot volume from the quality of opportunities. [1][2]
However, a total also reflects how much the player has played and the role they occupy. Someone who takes penalties is not directly comparable with someone whose total excludes them. Check the source's definitions before drawing a conclusion.
Goals minus xG describes a difference between observed scoring and the model's expectation. It does not, on its own, establish permanent finishing ability or prove that a player will keep scoring at the same rate.
Likewise, a player's shooting xG does not capture every contribution. Creating space, progressing the ball or making a pass before another player's shot requires other evidence. Use xG to answer a question about shots, rather than asking it to summarize the entire player.
Why sample size and match context matter
A single attempt is either converted or not, while its xG value is a probability. Over a short period, a few outcomes can therefore create a substantial difference between goals and xG. [1][2]
Larger, comparable samples can make interpretation more useful, but there is no universal number of matches after which an xG comparison becomes conclusive. Opponents, playing time, role, data coverage and model quality still matter.
The sequence of a match matters too. A rebound attempt can only occur if the earlier play leaves a rebound available. A goal can change how both teams approach the rest of the game. An xG total describes the attempts that occurred, not every alternative sequence the match might have taken.
Look for context and patterns rather than treating a single match total as a verdict or a forecast.
Why do different providers report different xG values?
Two providers can evaluate the same shot differently without either number being a universal answer. Their recorded features, historical samples and modeling choices may differ. [1][2]
One model may capture details about defensive pressure or goalkeeper position that another does not. Providers can also revise their methods over time.
For meaningful comparisons:
- Use figures from the same source where possible.
- Check whether penalties and other shot categories are included.
- Check whether the model or coverage has changed between periods.
- Distinguish ordinary xG from post-shot measures.
Do not combine values from different providers and assume they were calculated on an identical basis.
What xG cannot tell you
xG is valuable context, but it cannot establish everything a reader might want to know.
- It cannot guarantee a score or result. Chance quality and match outcomes are different questions.
- It is not a complete measure of player ability. It concerns shots, not every action or contribution.
- It does not value every dangerous attack. Standard shot-based xG has no shot to evaluate when a promising move ends without an attempt.
- It cannot explain every miss or save. The estimate depends on the information the model records.
- It is not a ready-made forecast. Predicting future matches requires more than reading one xG total.
A pre-shot estimate also does not directly assess where the ball actually travels after contact. That is where post-shot information becomes useful. [1][3]
What is post-shot xG?
Standard xG evaluates an opportunity using information available before or at the moment of the attempt. Post-shot expected goals, often shortened to PSxG, adds information about the shot after it has been taken. [1][3]
A chance can look difficult before contact but become a demanding save if the shooter places the ball well. Conversely, a promising opportunity may produce an easy effort for the goalkeeper. Pre-shot and post-shot estimates answer different questions.
The StatsBomb example
StatsBomb's primary methodological explanation describes a post-shot model using information such as trajectory, speed and the estimated position where the ball would enter the goal. The model described there is trained on shots that are saved or scored; blocked and off-target attempts are assigned zero post-shot scoring probability. [3]
This helps explain why post-shot measures are used when assessing the difficulty of shots a goalkeeper faces. It does not establish that every provider uses identical inputs or treatment of shot categories.
The detailed methodology in that source describes a particular StatsBomb model, not a universal or necessarily unchanged specification for all present-day PSxG data. Always check the documentation attached to the figures you use.
Sources and further reading
- Hudl StatsBomb, “Expected Goals (xG) Explained.” Definition, common inputs, provider variation and overview of post-shot measures. Read the explanation. The originally supplied StatsBomb explainer URL redirects to this page.
- Opta Analyst, “What Is Expected Goals (xG)?” Probability interpretation, Opta model inputs, team/player interpretation and misconceptions about match results and overperformance. Read the explanation.
- Hudl StatsBomb, “A New Way to Measure Keepers' Shot Stopping: Post-Shot Expected Goals.” Primary methodological explanation of the post-shot model described above. Read the methodology.
For related SportQuantix reading, start with Sports Analytics: Understanding the Game Through Data. To see how a different sport uses a different performance measure, read NBA Offensive Rating Explained. Offensive rating and xG are not interchangeable metrics.
Understand sports through data
SportQuantix is a sports intelligence platform in development. This guide is educational; it does not claim that SportQuantix currently provides xG statistics or proprietary xG models.
Pre-register for a free account to hear when the platform goes live. Your account stays pending until launch.