Picture a winger reaching the byline and cutting the ball back to a striker eight metres out, who sweeps it first time towards the near post, only for the goalkeeper to scramble it wide. Seconds later the broadcast graphic shows something like 0.38 xG, and the commentator calls it a big chance missed.

That number did not come from anyone's opinion of the move. It was produced by a model that compares the shot with nearly a million earlier attempts, and following its route from the pitch to the screen explains what expected goals can and cannot say about a match.

From Pass to Probability

Every xG value is built in the same sequence, whether the shot is a tap-in or a hopeful effort from 30 metres. Data providers log the action, describe it in numbers and then ask a trained model how often comparable shots have ended in goals:

• Logging the event. Analysts or tracking systems record the shot, the player, the body part used and the action that came immediately before it.


• Describing the situation. Opta's model draws on more than 20 variables, from distance and angle to defensive pressure and the goalkeeper's position.


• Comparing with history. The model, trained on nearly one million shots from 40 competitions between 2018-19 and 2021-22, estimates how often similar attempts were scored.


• Assigning a value. The result sits between 0 for an impossible chance and 1 for a certain goal, so 0.1 means roughly one goal from ten such shots.


• Adding it to the match total. Each value joins the team's running tally, which becomes the xG figure shown at full time.

Because the inputs describe the chance and not the finish, a shot blazed over the bar and a shot saved on the line can carry exactly the same value.

Why the Number Is an Expected Value

The word expected comes from expected value, the statistical idea of an average outcome over many repeats, so no player can ever score 0.38 of a goal. Over a long run of similar chances the goals should land close to the xG total, but any single shot ends as either a goal or a miss.

That also means overperformance is banked, not owed. Opta's analysts point out that a striker who has scored more than his xG is not due a cold spell, and expecting the numbers to balance out is a version of the gambler's fallacy.

Betting companies were among the first to adopt xG, and the same expected-value thinking carries over to casino offers. At an online casino, a stay casino promo code is worth what its terms allow, and that comes down to the wagering requirement, the games that count towards it and how long the bonus stays valid, far more than the headline amount.

What the Model Weighs

Not every model looks at the same inputs, but the most influential factors are shared across providers. Some describe the shot itself, such as where it was taken and with which part of the body, while others describe the moment around it, from the pass that created it to the defenders in the way. Richer models add more of that context, and their numbers shift accordingly.

Where the Shot Comes From

Distance and angle carry the most weight in almost every model, because central shots close to goal are scored far more often than attempts from tight angles or long range. Body part matters too, and a header from a given spot converts less often than a shot with the foot from the same position.

The type of assist adds another layer to the calculation. StatsBomb's analysts note that chances from crosses are harder to convert than those created by ground passes, through balls or a dribble.

What Surrounds the Shooter

Context around the shooter is where providers differ the most. Opta's model includes defensive pressure, the clarity of the path to goal and an estimate of the goalkeeper's position, while Hudl StatsBomb also records shot impact height and the positions of every attacker and defender in frame.

Those extra details can move a value sharply, and StatsBomb gives the example of a chance rated 0.30 by a standard model rising to 0.65 once the goalkeeper is recorded as out of position. The same shot can therefore carry different values depending on which provider is quoted.

From One Shot to a Match Total

A team's match xG is usually presented as the sum of its shots, but shots within the same attack cannot all become goals. If a striker scores with the first attempt, the rebound never happens, so providers such as FBref combine shots in a single possession as probabilities.

FBref's own example involves one Schalke attack with three shots rated 0.41, 0.47 and 0.79, which would add up to 1.67. Treated as one possession, the chance of at least one goal works out at about 0.93, because the probability of all three missing is 0.59 � 0.53 � 0.21.

Across a full match Opta records around 25 shots on average, while most leagues produce only 2.5 to 3 goals per game. That gap is the reason xG exists, since shots are frequent enough to measure while goals are too rare to judge a single performance on their own.

Same Shots, Different Stories

Two players can take exactly the same number of shots and produce very different xG totals, because the model rewards the quality of chances, not the volume. Opta's 2022-23 data offers a clean example from two very different shooters, who each took 63 non-penalty shots that season.

Both players finished slightly above their xG, but S�nchez was shooting from much closer to goal, so the model expected more than three and a half times as many goals from the same volume. Penalties are left out of comparisons like this because providers give them a fixed value: 0.79 at Opta, 0.78 at Hudl StatsBomb since 2022 and 0.76 in many other models.

What xG Leaves Out

The most common complaint is that xG only sees shots. A dangerous cross that flashes across the six-yard box without a touch earns nothing, so a side that keeps creating near misses can look less threatening on paper than it felt in the stadium.

Finishing and goalkeeping also sit outside the basic number, since pre-shot xG describes an average player taking the chance. Post-shot xG covers some of that ground by valuing on-target shots once placement is known, and it is mainly used to judge goalkeepers.

Defenders of the metric argue that these gaps shrink over time. A replication study by American Soccer Analysis found xG ratio was still the best predictor of future points, though only after about eight matches in Europe's top five leagues, and its margin over simpler measures was smaller than earlier research suggested.

Models also disagree with one another more than broadcast graphics imply. FBref notes that a single Diogo Jota chance against Southampton was rated 0.68 by an older model and 0.90 by Opta's current one.

The Final Whistle on Expected Goals

Expected goals turns every shot into a probability, adds those probabilities together and shows which side created the better chances, whatever the scoreboard says. Its value grows with the sample, so one match total is a clue about performance, while a run of eight or more games starts to say something about what comes next.

Read that way, xG is a tool for questioning a result, not for rewriting it. Anyone using it to shape bets is still dealing in probabilities, and Gambling Help Online offers free, confidential support if wagering on matches starts to feel like more than a hobby.

FAQ

What does 1.5 xG mean in a match?

It means the chances a team created would produce about one and a half goals on average if similar shots were taken many times. It describes the quality of those chances, not a prediction of the score.

Why do different websites show different xG?

Each provider builds its own model with different inputs, from basic shot location to the positions of defenders and the goalkeeper. The same chance can therefore receive noticeably different values, especially when the keeper is out of position.

What is the difference between xG and xGOT?

Standard xG values a chance before the shot is struck, while xGOT, also called post-shot xG, is calculated only for shots on target and accounts for placement. It is mostly used to measure goalkeeping.

Does a higher xG mean a team deserved to win?

Not necessarily, because xG measures the quality of chances, not the expected result. One clear opening can outweigh many hopeful shots, and finishing, goalkeeping and luck all shape the final score.