Learn

Expected goals (xG), explained

xG is a probability score for each shot. Treat it as chance quality, not a prediction.

What you will learn

xG quantifies the chance quality of each shot from 0 to 1.
Goals – xG (over/underperformance) is noisier than xG itself; demand a long sample.
Compare xG/90 within similar roles before declaring a striker hot or cold.

Each shot is scored by location, body part, defender pressure, and similar features against a historical baseline.

Summed across a season, xG estimates how many goals an average finisher would have scored from those chances.

Players who score above their xG are either elite finishers, lucky, or some mix — single seasons rarely separate the three.

Watch for repeated outperformance across multiple seasons before treating it as skill rather than noise.

Penalties skew xG: separate them when comparing strikers across teams.

Some xG models penalise long-range shots heavily; specialists like deep set-piece takers can look worse than they play.

It is one of the better predictors but still leaves a wide error band. Use it alongside minutes, role, team xG creation, and age curves.

Different models, training samples, and shot-context features. Always check the methodology before comparing across sources.