Expected points is one of the most powerful predictive tools in football analytics. It strips away the noise of individual match results and reveals which teams are genuinely performing well and which are surviving on luck.
How xPTS Works
For each match, calculate the probability of each outcome using the match xG:
- Use a Poisson model to convert home xG and away xG into scoreline probabilities
- Sum all home win scorelines for P(Home Win)
- Sum all draw scorelines for P(Draw)
- Sum all away win scorelines for P(Away Win)
- Calculate xPTS: P(Win) × 3 + P(Draw) × 1
Example: Liverpool create 2.3 xG vs Newcastle's 0.8 xG
- P(Liverpool Win) = 72%, P(Draw) = 18%, P(Liverpool Loss) = 10%
- Liverpool xPTS = (0.72 × 3) + (0.18 × 1) = 2.34
Sum across all matches for the season total.
Identifying Overperformers and Underperformers
Overperformers (Actual Points > xPTS)
These teams have been converting chances at above-average rates, winning tight matches, or benefiting from opponent errors. Their results will likely worsen as regression kicks in.
Betting angle: Bet against overperformers in upcoming matches, particularly when their odds are short based on their inflated league position.
Underperformers (Actual Points < xPTS)
These teams are creating good chances but not converting, or conceding goals from low-quality chances. Their results will likely improve.
Betting angle: Back underperformers, especially at home where odds may have drifted due to poor recent results.
xPTS vs Actual Table: A Case Study
Consider a mid-season snapshot where Team A sits 5th with 35 points from 20 matches but has only 28 xPTS. Team B sits 12th with 25 points but has 32 xPTS.
The xPTS table says Team B is the genuinely stronger side. Over the remaining 18 matches, Team B is likely to gain ground while Team A drops. Backing Team B at inflated odds — and opposing Team A at compressed odds — is a classic xPTS-driven strategy.
Limitations of xPTS
Not All Overperformance Is Luck
Teams with elite goalkeepers (who outperform xGA consistently) or clinical strikers (who outperform xG consistently) can sustain modest overperformance. A gap of 3-4 points over a full season may reflect genuine quality not captured by the xG model.
Sample Size Matters
xPTS is unreliable with fewer than 10 matches. The model needs sufficient data to produce stable estimates. Do not draw conclusions from 5-match xPTS gaps.