xG (Expected Goals) measures the quality of scoring chances. Each shot gets a value between 0 and 1 based on where it was taken, how it was created, and the circumstances around it: a penalty carries an xG of around 0.79, a long-range effort might be 0.03. Add up a team’s chances and you have their xG for the match.
The power of xG is that it separates performance from results. A team can win 3-1 having created chances worth 0.8 xG while their opponent created 2.1; the scoreline says dominance, the chance quality says the opposite. Final scores tell you what happened. xG tells you what the chances created deserved.
Shot counts mislead for the same reason. Fifteen shots from distance are worth less than five from inside the six-yard box, and a match that looks one-sided on shot volume can be level or worse on chance quality. xG prices every shot rather than counting them.
In the model, xG is raw material. Pre-match, attacking and defensive ratings built on xG feed the goal expectancies that drive the Poisson scoreline grid and every step after it. In-play, live xG accumulates as chances arrive and drives the goal-pressure read and the overdue calculation. The xG summary on your analysis shows the full-match and half-by-half figures the probabilities were built from.
Further reading on the blog: The Ultimate Guide to xG Stats and xG vs Shot Volume: Why Shot Counts Lie.