How it works
Take a match with xG 2.1 vs 0.9. Simulating goals from those chance qualities, the better side wins in most scenarios — say 62% win, 22% draw: xPts = 0.62 × 3 + 0.22 × 1 ≈ 2.0 against 1.0-ish for the opponent. Summed over a season, xPts build the «deserved» league table.
What it is for
Comparing real points with xPts exposes over- and under-performers: a team 8 points above its xPts is riding variance, one far below is better than the table shows — often the more interesting betting signal.