Expected Points: How AI Builds a Truer League Table

The official league table tells you how many points teams have actually accumulated, but it does not tell you how many points they deserved based on their underlying performance. Expected Points (xPoints) tables, powered by AI analysis, provide this deeper insight and are a powerful tool for identifying teams likely to improve or decline. Our models at 1X2.TV use xPoints as a key input for match predictions.

What Are Expected Points?

Expected Points are calculated by simulating each match thousands of times using the expected goals (xG) created and conceded by each team. If Team A created 2.0 xG and conceded 0.8 xG in a match they drew 0-0, the xPoints model would show they deserved approximately 2.3 expected points from that match rather than the 1 point they actually earned. Aggregated across all matches, xPoints reveals which teams are over- or underperforming relative to their underlying quality.

Overperformers and Underperformers

Overperforming Teams

Teams with significantly more actual points than expected points are benefiting from some combination of: clinical finishing (converting a high percentage of chances), luck in close matches, strong set-piece execution, or exceptional goalkeeping. While these factors are partially skill-based, extreme overperformance tends to regress toward xPoints over time, making these teams candidates for future decline.

Underperforming Teams

Conversely, teams with fewer actual points than expected are creating good chances but not converting, losing close matches to bad luck, or conceding from low-xG opportunities. These teams are strong candidates for improvement, and our models increase their predicted performance relative to their actual league position.

Using xPoints in Predictions

Our AI models at 1X2.TV weight xPoints data alongside actual results, creating a blended view that captures both demonstrated results and underlying performance quality. This approach produces more accurate long-term predictions than using either metric in isolation.