xG for, xG against, and regression

One match of xG is interesting; a rolling window of xG is where prediction lives. Splitting it into the two sides of the ball — and watching how teams drift back to it — is the heart of reading form.

xG for vs xG against

• xG for is the quality of chances a team creates — its attacking output.
• xG against is the quality of chances it allows — its defensive solidity.

The gap between them (xG difference) is one of the cleanest summaries of how good a team really is. A side averaging 1.8 xG for and 0.7 xG against is dominating both ends; one at 1.0 for and 1.6 against is being out-chanced even if results have been kind.

Overperformance and underperformance

Compare a team's actual goals to its xG and you spot luck and finishing:

• Scoring well above xG (overperforming) usually means hot finishing or a clinical striker — and it tends to cool off.
• Scoring below xG (underperforming) often means wasteful finishing or bad variance — and it tends to recover.

Regression to the mean

Both gaps shrink over time. A team riding a 7-goals-from-3-xG streak is very unlikely to keep it up; the smart read expects them to drop back toward their underlying xG. This is regression to the mean, and it is why a model trusts repeatable chance creation over a recent flurry of goals.

Results lie in the short run; underlying xG tells you where a team is heading. Bet the process, not the last scoreline.

For predictions, an xG-flavoured form window is far harder to fool than raw goals — exactly why FinalSkore leans on it rather than the bare scoreline.