What is expected goals (xG) in football?
知识库 · 发布于 · 阅读约3分钟
本文仅提供英文版本。
Expected goals, or xG, is a number between 0 and 1 attached to a shot. It estimates how likely that kind of chance is to end in a goal, based on how similar shots have turned out in the past. Add up the values for every shot a team takes and you get a rough measure of the quality of the chances it created, not just the final score.
How xG values a shot
A model looks at features of each shot, such as distance, angle, body part, and the type of pass or situation that created it. A shot with an xG of 0.05 is the kind that scores about one time in twenty. A shot with 0.50 is the kind that scores about one time in two. A penalty is rated by most models at roughly three in four, though the exact figure depends on the model.
Because each provider builds its own model with its own data and inputs, the same shot can get slightly different values from different sources. Compare numbers from one provider with numbers from the same provider.
A worked example
Suppose Team A takes four shots in a match, valued by a model at 0.05, 0.12, 0.31 and a penalty at 0.76.
- Team A xG = 0.05 + 0.12 + 0.31 + 0.76 = 1.24
Team B takes four shots valued at 0.08, 0.06, 0.15 and 0.40.
- Team B xG = 0.08 + 0.06 + 0.15 + 0.40 = 0.69
The xG difference is 1.24 - 0.69 = 0.55 in favour of Team A. The real score might still be 0-0 or 1-2. xG describes the chances, not the result, so a team can win while creating fewer good chances, or lose despite creating more.
What xG does and does not tell you
xG is useful for judging whether a result was fair, spotting teams whose scoring runs look lucky or unlucky, and seeing the quality of attacks and defences over several matches. A single match is a small sample, so one game says little. Over many matches the picture is steadier.
It does not capture everything. Many models ignore the position of defenders, the goalkeeper's skill or the finisher's ability, and they say nothing about chances that were never recorded as a shot. It is one input, not a verdict. For other context, see how to read head-to-head records and form.
FAQ
How is xG calculated?
A statistical model is trained on large numbers of past shots. For a new shot it compares features such as distance, angle and shot type against similar past shots and returns the share that ended up as goals. The sum of these values across a match is the team's xG.
Is xG a good predictor?
It is a useful input, and over longer spans it often reflects underlying performance better than goals alone, because goals are rare and noisy. It is not a prediction of the next score, and it does not guarantee anything about the next match. Treat it as one piece of information alongside lineups, injuries and form.
What is xG difference?
It is a team's xG minus the xG it conceded, or the gap between two teams' xG in a match. In the example above it is 0.55. A positive number means the team created better chances than it allowed.
Why do different sites show different xG numbers?
Each provider uses its own model, data and inputs, so values for the same match can differ. Stick with one source when comparing teams.
See also
- Football fixtures and pages
- Football predictions
- Football trends
- Expected value calculator
- How to read head-to-head and form
- What is expected value?
Betting is for adults only (18+). Only stake money you can afford to lose, set limits before you start, and stop if it stops being fun. If gambling is causing you problems, contact a local support organisation.