## The xPoints Concept
Expected Points (xPoints) calculates how many league points a team should have earned based on the xG values of all chances in all their matches, rather than actual results.
If your xG model predicts each match result probability, you can calculate the expected points from those probabilities:
xPoints per match = P(win) × 3 + P(draw) × 1 + P(lose) × 0
Summing xPoints over a season gives the expected league standing based on underlying performance — independent of results variance.
## The Over- and Under-Achievers
Teams that significantly outperform xPoints have been lucky with results relative to their underlying performance. Teams that significantly underperform xPoints have been unlucky.
**Historical research finding:** The correlation between points over/under-performance in one half of a season and points performance in the second half is strongly negative. Over-performers in H1 tend to under-perform in H2. Under-performers in H1 tend to over-perform in H2.
This is regression to the mean in league table form.
## Identifying the Regression Candidates
After 15–20 matches of a season:
1. Calculate each team's actual points
2. Calculate their xPoints from your model
3. Identify the largest positive and negative divergences
Teams with actual points significantly above xPoints: potential value to back their opponents in subsequent matches (or lay them on the exchange for their outright position).
Teams with actual points significantly below xPoints: potential value to back them in subsequent matches at inflated prices.
## The Season-Position Outright Market
The xPoints divergence is most directly exploited in outright markets: a team priced short for the title or a Champions League position based on actual results, when xPoints suggests their position is significantly above their underlying quality, may offer lay value.
Create a free account to track your progress and save bookmarks.
