xG against reality: why the gap is instructive
Expected goals measure the quality of chances, not the score. The gap between the two says something, provided you know over how many matches to read it.
By The PRIZIFOOT team
xG (expected goals) assigns each shot a probability of ending up in the net, estimated from comparable shots: distance, angle, body part, phase of play. Summed over a match, it gives the goals a team “should have” scored given its chances.
Why the gap is not pure noise
A team scoring well above its xG over three matches has probably been lucky. The same team over sixty matches may simply have better finishers. The difficulty is that both situations produce exactly the same chart on a short sample.
That is why a gap must always be read with its sample size beside it. A figure without its denominator is not information.
The careful use
xG is better at correcting a reading than at producing one. A team on a run of flattering results but a mediocre xG is a candidate for regression; the reverse holds too. It is one signal among others, not a prediction.
Move from theory to the match
Pick two teams and see what the model makes of it, without creating an account.
Analyse a match