What xG measures
An expected-goals model assigns each shot a scoring probability using historically similar attempts. A 0.20 xG shot means that this type of chance is converted about once in five attempts; it does not promise one goal after exactly five shots.
Team xG adds those probabilities. A side can score twice from 0.8 xG in one match through excellent finishing or ordinary variance, so stronger conclusions need a run of matches rather than a single final score.
- xG describes chance quality, not every action in the attack.
- Models use different inputs, so compare figures from the same provider.
Read shot quality before the total
Location, angle, assist type, defensive pressure and body part can all change shot quality. A close-range attempt from a low cut-back is not equivalent to a long shot into a set defence, even though both count as one attempt.
Ask how the total was built. Did 1.2 xG come from one huge chance plus weak shots, or six medium opportunities? The distribution often says more about repeatability than the headline number.
- Separate big chances from low-probability volume.
- Check how the data provider treats blocked shots and set pieces.
Game state changes behaviour
A leading team may defend deeper and concede possession or distant shots, while a trailing team commits more players and accepts more risk. A late xG surge after falling behind therefore does not prove that the chasing side controlled the entire match.
Split the match into level, leading and trailing periods. Add red cards, substitutions and time remaining so that a contextual effect is not mistaken for stable team quality.
- Compare level-score xG to understand the base plan.
- Place red cards, goals and minutes beside the shot timeline.
A quick worked reading
Suppose Team A finishes with 1.6 xG to 1.1, but creates 1.0 after going two goals down. The fair conclusion is not complete control. Team B may have won the level phase before allowing late pressure while protecting the score.
Start with the shot map, inspect the three largest chances, then align their timing with score changes. Finally, compare the pattern across several matches before turning one performance into a verdict.
- Ask when and how the chances were created.
- Use xG as a diagnostic tool, not a replacement scoreboard.