Goals are few and luck affects them: a post, a deflection, a keeper having the night of his life. xGi (expected goals) tries to measure what a team controls, the chances it creates.
How it is computed
A statistical model estimates how often a shot from a given spot, in a given way, ends in a goal. If 3 in 10 from a spot go in, a shot from there has an xG of 0.30.
How to read it
- If a team consistently has more xG than its opponents, it plays better than the scores show.
- If it scores far above its xG for many matches, it either has outstanding finishers or luck. Usually it is the second, and the gap tends to close.
- A team's xG over 3 matches says little. Over 10 to 15 matches it starts to support a conclusion.
Example
Team A wins 3-0 with xG 0.6 against 2.4. Team B wins 1-0 with xG 2.4 against 0.6. B created the better chances. In the next match the model will expect more from B than from A.
Its limits
xG does not know who is shooting, nor how much pressure the player is under from defenders, unless the model records it. It does not measure good defending that prevents shots either. It is a useful tool, not the whole picture.
xG becomes the average goals that feed the Poissoni model: Poisson and Dixon-Coles.