Expected Points (xPts) — How to Turn xG into League Table Predictions

July 8, 2026 · 12 min read

Expected Goals (xG) tells you how well a team played in a single match. Expected Points (xPts) tells you how well a team should be doing across an entire season. If you want to predict league standings — not just individual results — xPts is the metric that bridges the gap between match-level analytics and season-long forecasting.

What Is Expected Points (xPts)?

Expected Points takes the xG from a single match and converts it into the number of points a team "should" have earned based on the quality of chances created and conceded. Where xG tells you "this team created 2.3 expected goals worth of chances," xPts tells you "this team would earn 2.1 points from this match on average."

The concept is straightforward: if a team consistently creates far more xG than they concede, they are generating winning chances. Over 38 matches, that pattern should translate into points and league position. xPts quantifies exactly how many points.

The key insight is that xPts is not just "xG converted to points." It is a probabilistic model that accounts for the full range of possible scorelines in each match, weighted by their likelihood. A team with 2.0 xG does not simply "deserve 2 goals" — they deserve a probability distribution across 0, 1, 2, 3+ goals, and each outcome yields different points.

The Poisson Method: How xPts Is Calculated

The standard approach to calculating xPts uses the Poisson distribution — a statistical model that predicts the probability of a given number of events occurring in a fixed interval. In football, the "events" are goals, and the "interval" is a 90-minute match.

The Poisson distribution is defined by a single parameter λ (lambda), which represents the expected number of goals. In xPts calculations, each team's xG from a match serves as their λ value. The formula for the probability of scoring exactly k goals is:

P(X = k) = (λ^k × e^−λ) / k!

To calculate xPts for a single match, the model follows four steps:

1. Set λ values: Team A's λ = their xG in the match. Team B's λ = their xG. For example, if Team A created 1.8 xG and Team B created 0.9 xG, then λ_A = 1.8 and λ_B = 0.9.
2. Build the scoreline matrix: Calculate P(Team A scores i goals) × P(Team B scores j goals) for every combination of i and j (typically 0 through 6, since probabilities beyond that are negligible).
3. Aggregate outcome probabilities: Sum all scorelines where Team A wins to get P(Win), sum all draws for P(Draw), and sum all scorelines where Team B wins for P(Loss).
4. Calculate expected points: xPts = 3 × P(Win) + 1 × P(Draw) + 0 × P(Loss). This gives the points a team would earn on average from this match if it were played repeatedly with the same chances.

Over a full season, a team's total xPts is simply the sum of their xPts from each individual match. This gives you a season-long expected points total that can be compared directly against the actual league table.

Real-World Data: Six Seasons of EPL xPts vs Actual Points

The most compelling way to understand xPts is to see it in action. Understat tracks xPts for every match across major European leagues. Here is what six seasons of Premier League data reveal about the gap between what teams deserved and what they actually earned.

The Biggest Over-Performers (Actual Points Far Exceed xPts)

These teams earned far more points than their shot quality suggested. Over a full season, they consistently converted low-quality chances into goals or won matches they "should" have drawn or lost.

The Biggest Under-Performers (xPts Far Exceed Actual Points)

These teams created enough chances to earn significantly more points but could not convert. They hit the woodwork, missed penalties, conceded late goals from low-xG chances, or simply could not finish.

Brighton's 2020/21 season is the most extreme case in Premier League history. They generated 61.41 xPts — enough for a comfortable mid-table finish — but ended with just 41 points and were dragged into a relegation battle. Their xG suggested a team finishing 10th; the actual table had them 16th.

Why Do Teams Over- or Under-Perform xPts?

The gap between xPts and actual points is not random noise. Research and six seasons of data point to several systematic factors:

• Finishing quality: Some teams have elite finishers who consistently beat xG. Erling Haaland, for example, regularly outperforms his individual xG by 20–30%. If a team has multiple clinical finishers, they will over-perform their xPts.
• Goalkeeper performance: A goalkeeper who saves 5–10% more shots than expected (measured by post-shot xG, or xGOT) can swing 5–8 points across a season. Emiliano Martínez's 2022/23 campaign at Aston Villa was a textbook example.
• Game state management: Teams that protect leads effectively — even with low xG — accumulate points beyond what their chances suggest. José Mourinho's Tottenham in 2020/21 earned 62 points from just 53.98 xPts by winning tight matches.
• Set-piece efficiency: xG models underweight set-piece quality. Teams with elite set-piece coaching (like Brentford or Arsenal under set-piece coach Nicolas Jover) consistently outperform xPts because their set pieces generate higher-quality chances than the model assumes.
• Regression to the mean: Over- and under-performance tends to correct over time. Brighton's −20.41 in 2020/21 was followed by seasons much closer to their xPts. This is the single most important insight for prediction: extreme deviations are not sustainable.

How to Use xPts for Predictions

xPts is most powerful when you use it to identify teams whose current league position is misleading. Here is a practical framework:

Step 1: Compare xPts Ranking to Actual Ranking

Pull xPts data from Understat or FBRef for the current season. Compare each team's xPts ranking to their actual league position. A team sitting 5th in xPts but 12th in the actual table is likely to climb. A team sitting 3rd in the table but 10th in xPts is likely to fall.

Step 2: Check the Size of the Gap

Small gaps (±3 points) are normal variance. Medium gaps (±5–8 points) suggest a pattern worth monitoring. Large gaps (±10+ points) almost always indicate unsustainable performance. In the six seasons of EPL data, every team with a gap of +10 or more saw significant regression the following season.

Step 3: Identify the Cause

Before acting on an xPts gap, determine why it exists. If a team is over-performing because of an elite goalkeeper who is saving everything, that might sustain longer than a team over-performing because of a striker on an unsustainable hot streak. Context matters.

Step 4: Apply to Match Predictions

When predicting individual matches, use xPts-adjusted rankings rather than actual league position. A team that is 15th in the table but 8th in xPts is playing much better than their results suggest. Betting markets and casual fans anchor on the actual table — xPts gives you an edge by revealing the underlying performance.

xPts vs xG: What Each Metric Tells You

It is worth clarifying the distinction between these two metrics, because they are often confused:

• xG (Expected Goals): Tells you the quality of chances in a single match. "This team created 2.3 xG" means they had chances that, on average, would produce 2.3 goals. It is a match-level metric.
• xPts (Expected Points): Tells you how many points a team deserved from a match or season, based on the balance of chances. It converts xG into a win/draw/loss probability and then into points. It is a predictive metric designed for league tables.
• xGD (Expected Goal Difference): Tells you the gap between chances created and conceded. xGD correlates strongly with league position over a season — stronger than actual goal difference in many cases.

The relationship is: xG feeds into xPts through the Poisson model, and xPts aggregates across matches to give a season-level picture. xG is the input; xPts is the output. You need both, but they answer different questions.

Limitations of xPts

xPts is a powerful tool, but it has real limitations that every user should understand:

• It ignores game state: xPts treats all chances equally, regardless of when they occur. A team creating 1.5 xG while chasing a 1-0 deficit plays differently than a team creating 1.5 xG while protecting a 1-0 lead. xPts does not capture this.
• It assumes independence between goals: The Poisson model assumes each goal is an independent event. In reality, goals change match dynamics — a team that scores first may sit back, reducing both teams' subsequent xG.
• xG providers differ: Understat, FBRef, Opta, and StatsBomb all calculate xG differently. Their xPts values for the same team in the same season can vary by 3–5 points. Always use the same provider for comparisons.
• It does not account for red cards, penalties won, or own goals: These events significantly affect match outcomes but are often excluded from or poorly modeled in xG calculations.
• Small samples are noisy: After 10 matches, xPts can be heavily influenced by one or two outlier performances. It becomes more reliable after 20+ matches, and most predictive after a full 38-match season.

The Brighton Paradox: A Case Study

Brighton's 2020/21 season remains the most famous xPts case study in Premier League history. Under manager Graham Potter, Brighton played attractive, possession-based football that consistently created high-quality chances. Their xG and xPts numbers were excellent — they generated 61.41 xPts across 38 matches, which would have placed them comfortably in 10th.

But they finished with 41 points, just 16th, and spent much of the season in a relegation battle. The gap of −20.41 points is the largest single-season under-performance in the xPts era.

What went wrong? Brighton's finishing was abysmal. They converted chances at a rate far below what xG expected, missing clear-cut opportunities repeatedly. Their strikers — Neal Maupay and Danny Welbeck — underperformed their individual xG by significant margins. Meanwhile, opponents were clinical: teams scored from low-xG chances against Brighton at an above-average rate.

The following season (2021/22), Brighton's xPts and actual points converged. They finished 9th with 51 points, much closer to their underlying performance level. The Brighton Paradox — great process, terrible results — corrected itself, as xPts theory predicted it would.

How FanPick Uses xPts Thinking

On FanPick, understanding xPts can sharpen your predictions. When you see a team that has been winning ugly — scraping 1-0 victories despite being outshot — their xPts will be lower than their actual points. That is a team likely to regress. Conversely, a team losing close matches despite dominating chances is due for a positive run.

Use xPts data to identify these regression candidates before the market catches up. The prediction edge comes from recognizing that past results are not always the best predictor of future performance — underlying chance quality is.

Key Takeaways

• xPts converts match-level xG into expected points using the Poisson distribution, giving you a season-long view of team quality independent of finishing luck.
• Extreme xPts gaps always regress. In six EPL seasons, no team sustained a +15 or −15 point gap for two consecutive seasons.
• Use xPts to spot regression candidates. Teams outperforming xPts by 10+ points are likely to drop. Teams underperforming by 10+ points are likely to rise.
• xPts is not perfect. It ignores game state, assumes independent goals, and varies by provider. Use it as one input among several, not as gospel.
• Context explains the gap. Before acting on an xPts discrepancy, identify whether it is driven by finishing quality, goalkeeping, set pieces, or pure variance — each has different sustainability.