Expected goals is the most useful football metric of the last decade and the most misread. Here is what it does and what it does not.
What xG measures
Every shot gets a value between zero and one, the probability it is scored from that position and situation. Add them up and you have expected goals for a match. A side with 2.4 xG created better chances than a side with 0.7 xG, whatever the scoreline said.
Why it beats the scoreline
Goals are rare and noisy. A deflection or a worldie can decide a match that one side dominated. xG smooths that noise and tells you who created the better chances, which is a stronger signal of repeatable performance.
Where xG misleads
xG is not destiny. It does not know game state, red cards or a keeper having the night of his life. A single match is a small sample, so one xG figure proves little. The value comes over many matches, where finishing luck evens out and quality shows.
How we use it
In the baseline, an expected-goals proxy is one of three inputs, never the whole story. It sits alongside Elo and form, and the model shows you how much each contributed.
See expected goals and shot maps in the glossary.
