What Is Expected Points (xPts)? Calculating How Many Points a Match "Should" Yield from xGGuides · Football · Stats explained

What Is Expected Points (xPts)? Calculating How Many Points a Match "Should" Yield from xG

Key points

  • xPts measures how many points should be earned on average based on match performance
  • Formula: 3 × Win Probability + 1 × Draw Probability
  • Best suited for checking whether a team's points tally is unwarranted

1What Is xPts: How Many Points Should Be Earned Based on Performance

Expected Goals (xG) tells you how many chances both sides created in a match. Expected Points (xPts) takes it a step further: based on those chances, how many points would be earned in this match on average?

Formula: xPts = 3 × Win Probability + 1 × Draw Probability

Example

In a match, the home team has 1.8 xG and the away team has 0.9 xG. Converted, the home team's win probability is 58.6%, draw is 22.9%, and loss is 18.5%:

TeamWinDrawCalculationxPts
Home team58.6%22.9%3 × 0.586 + 0.2291.99
Away team18.5%22.9%3 × 0.185 + 0.2290.78

If the actual match ends 0:1 with the away team winning, the home team gets 0 points and the away team gets 3 points. The home team "underperformed" by 1.99 points, and the away team "overperformed" by 2.22 points. Accumulated over a 38-match season like this, it becomes clear which teams have inflated points tallies and which teams were undermined by luck.

The xPts of both teams adds up to 2.77, not 3. This is because in a draw both teams get 1 point each, totaling only 2 points; the higher the probability of a draw, the lower the sum.

Why not look directly at xG difference? Because league tables count points, not goals. Winning 3:0 on xG versus winning 1.2:1.0 shows a huge difference in xG difference, but on the points table, both are worth only one win; xPts translates "chances" into the same language as the league table, allowing for a direct comparison with actual points.

2How It Accumulates Across a Season: Demonstrated with 3 Matches

xPts is calculated match by match, then totaled. Suppose a team plays 3 consecutive matches:

MatchxG (This team vs Opponent)Actual ScoreActual PointsxPts
Match 11.8 vs 0.92:1 Win31.99
Match 21.0 vs 1.00:1 Loss01.35
Match 30.6 vs 2.11:1 Draw10.46
Total——43.80

In Match 2, performance was 50-50 yet the team lost, earning 1.35 fewer points; in Match 3, the team was dominated yet snatched 1 point, earning 0.54 more points; in Match 1, winning yielded 1.01 points more than expected. Across the 3 matches combined, actual points totaled 4 and xPts was 3.80; good luck and bad luck canceled each other out, with a difference of only 0.2 points.

Single-match discrepancies can be large, but over a season they mostly cancel out. If by the end of the season the gap is still more than 10 points, it is unlikely to be pure luck and is worth investigating the causes.

3From xG to Win, Draw, and Loss: The Poisson Method

The most common approach is to treat xG as the average number of goals, using the Poisson distribution to calculate the probability of each goal count.

GoalsHome Team (xG 1.8)Away Team (xG 0.9)
016.5%40.7%
129.8%36.6%
226.8%16.5%
316.1%4.9%
47.2%1.1%
52.6%0.2%

Multiplying each team's probabilities together yields the probability of each scoreline:

ScorelineCalculationProbability
1:029.8% × 40.7%12.1%
1:129.8% × 36.6%10.9%
2:026.8% × 40.7%10.9%
2:126.8% × 36.6%9.8%
0:016.5% × 40.7%6.7%

Summing all scorelines where the home team scores more gives the win probability of 58.6%; summing scorelines where both teams score equally gives a draw probability of 22.9%.

Behind this algorithm lies an assumption: the goals of both teams are independent of each other. In reality, the team that scores first often drops back to defend, while the trailing side presses forward, so real matches usually produce slightly more draws than Poisson calculates. It is sufficient for viewing a season's broad direction, but single-match details should not be taken too strictly. Manual calculations are tedious; plugging both teams' xG into the Poisson Score Calculator gives the results immediately.

4Comparison Table of Common xG Combinations

Calculating several combinations using the same method gives a feel for xG and xPts:

xG (Team A vs Team B)Team A WinDrawTeam B WinTeam A xPtsTeam B xPts
2.4 vs 0.677.4%15.2%7.4%2.470.37
1.8 vs 0.958.6%22.9%18.5%1.990.78
1.2 vs 1.138.3%28.3%33.4%1.431.28
1.0 vs 1.034.6%30.9%34.6%1.351.35
1.5 vs 1.537.8%24.3%37.8%1.381.38

A few patterns:

  • When the xG difference is around 1 goal, the xPts difference is approximately 1.2 points.
  • No matter how large the xG, it never reaches 3. Even with dominance like 2.4 versus 0.6, xPts is only 2.47, because football is low-scoring and there is always room for upsets.
  • Draws are most frequent when both sides are low. A 1.0 versus 1.0 matchup has a 30.9% draw probability, higher than the 24.3% of 1.5 versus 1.5.

The last point is very practical: in matches where both teams are conservative and chances are scarce, even if one team is slightly better, their xPts will be very close because the probabilities of 0:0 and 1:1 are very high. Conversely, in an open, attacking slugfest, a team with a slight edge in xG pulls away more significantly in win probability.

5One Big Chance vs. Many Small Chances

Poisson only looks at the total xG. A more granular approach is to treat each shot as an independent event, simulating shot by shot. At this point, one discovers: even with the same total xG, if the shape of the chances differs, the xPts will also differ.

Example: Both at 0.6 xG

Team A has only one one-on-one chance with 0.6 xG; Team B takes 6 shots, each with 0.1 xG.

Team A (1 × 0.6)Team B (6 × 0.1)
Probability of scoring 0 goals40%53.1%
If opponent scores 0: Win60%46.9%
If opponent scores 0: Draw40%53.1%
If opponent scores 0: xPts2.201.94

This is a simplified scenario: assuming the opponent does not score and comparing only the team's chance distribution, rather than a full match simulation. The probability of Team B missing all 6 shots is 0.9 to the 6th power, or 53.1%, which is higher than the 40% chance of Team A missing its single shot. Although Team B could potentially score 2 or more goals, that probability is very small; Team A has a greater chance of scoring at least 1 goal, so in this simplified scenario, its xPts is actually higher.

Translated into match terms: a single one-on-one chance against the goalkeeper is more likely to yield points than 6 long-range shots from outside the box, even if both sum to the same xG. This is also why managers would rather take fewer shots to wait for better opportunities.

Some websites use shot-by-shot simulations while others use Poisson, so for the same match, there may be a slight difference compared to calculating xPts yourself with Poisson.

62025-26 Premier League: Who Earned More Points Than Their Performance

Totaling each match's xPts across an entire season and comparing it against actual points: Understat calculates match xPts based on its own xG model, then aggregates them for the full season. Below are selected 2025-26 Premier League teams with notable discrepancies, ordered from largest to smallest difference:

TeamActual PointsxPtsDifference
Aston Villa6551.07+13.93
Sunderland5442.03+11.97
Fulham5245.08+6.92
Manchester United7164.45+6.55
Arsenal8579.87+5.13
Chelsea5258.85-6.85
Tottenham Hotspur4149.25-8.25
Crystal Palace4553.88-8.88
Leeds United4756.50-9.50
Wolverhampton Wanderers2035.44-15.44

Among the listed teams, three had a discrepancy of more than 10 points between actual points and xPts: Aston Villa, Sunderland, and Wolverhampton Wanderers. Aston Villa earned nearly 14 additional points; based on underlying performance they only warranted around 51 points, yet their points tally reached the top 4. Wolverhampton Wanderers underperformed by more than 15 points; based on chance quality, they should not have had only 20 points. Arsenal were champions, and their xPts was also among the highest tier, demonstrating that the title was not built on luck.

A positive "Difference" column indicates earning more points than underlying performance warranted; a negative value indicates performance was better than the points tally. Aston Villa also won the Europa League title that season, and the team's ability to capitalize on chances in critical moments was indeed very strong, but a 14-point gap from an xPts perspective is still considered exceptionally large.

Switching to another model yields different evaluations. Opta's expected points table in early March 2026 ranked Aston Villa 13th when they were actually in 4th place at the time, also concluding that Villa's league standing was higher than their underlying performance. These were mid-season figures and cannot be directly compared side-by-side with Understat's end-of-season figures. Different models evaluate each shot's xG differently, which leads to differences when aggregated.

7How to Use xPts

xPts is best suited to answer: "Does this team's points tally fail to reflect reality?" The following three scenarios can be seen in every league and every season, differing only in magnitude:

ScenarioCommon CausesTypical Trajectory Next Season
Points far higher than xPtsSuperb finishing, heroic goalkeeping, many narrow winsTends to correct downward
Points far lower than xPtsPoor shooting luck, frequently conceded late goalsTends to rebound if performance remains unchanged
Points close to xPtsPoints match realityDepends on squad changes

Example (Hypothetical): Two Teams After 10 Rounds

TeamActual PointsxPtsDifferenceHow to Interpret
Team A2214.5+7.5Strong standing, but performance is only mid-table
Team B1117.0-6.0Poor standing, but performance is actually not bad

Looking only at the league table, Team A is competing for the top 4 and Team B is in the relegation zone. Looking at xPts, Team B's match performance is better than Team A's. If both teams continue to perform over their upcoming games just as they did in the first 10 rounds, the points gap will typically narrow gradually. This is where xPts is most commonly used: distinguishing between "genuinely strong" and "running hot right now".

However, there are two things xPts cannot see:

  • Contracting after taking the lead. When a team drops back to defend after leading 1:0 and their opponent catches up in xG, xPts underestimates the fact that the team was actually in control of the match.
  • Elite finishers and goalkeepers. Truly exceptional strikers can consistently convert 0.1 xG chances over the long haul, which is not purely luck; elite goalkeepers can also deny high-quality opponent chances, keeping a team's goals conceded below expectation over the long run.

This follows the same concept as Pythagorean Expectation, which uses points/runs scored and allowed to calculate "expected record"; the difference is that xPts uses chance quality rather than actual goals scored. The league table for the current season is in the Premier League Hub.

8Frequently Asked Questions

How big of a gap between xPts and actual points is considered large?

Across a 38-match season, a difference within 5 points is mostly normal fluctuation; once the gap reaches 10 points or more, it is well worth looking back to see whether that team had exceptional finishing or extraordinary luck. With small samples mid-season, a difference of 5 or 6 points after 10 rounds is very common; trends only become more apparent by season's end.

Why do different websites show different xPts?

Two reasons: each provider's xG model assigns different values to the same shot; and when calculating xPts, some use Poisson while others simulate shot-by-shot. Therefore, comparisons should always use figures from the same source.

Which is more useful: xPts or xG difference?

For evaluating overall team quality, xG difference (xG minus conceded xG) is more stable because it does not depend on how matches end. If you want to compare against the league table to see which team's points tally is inflated, xPts is more direct because its unit is points. Looking at both together provides the most complete picture.

Can a single match's xPts exceed 3 or fall below 0?

No. xPts is "3 × Win Probability + Draw Probability"; at most it approaches 3, and at lowest it approaches 0.

In the 2026-27 season, who currently has more points than their performance warrants?

You can visit Understat's Premier League page, switch to the new season, and compare each team's actual points with xPts. The sample is very small early in the season, and a difference of 5 to 6 points after around 10 rounds is very common; take it as a reference for now and do not rush to conclusions.

Can I calculate it myself?

Yes. Enter both teams' xG into the Poisson Score Calculator to get the win, draw, and loss probabilities, then apply the formula: 3 × Win + 1 × Draw. If you want to calculate for an entire season, add up the results of each match, then subtract that from the actual points on the league table.

9Sources

Illustrated portrait of Old Parker

About the author

Old Parker

Football columnist

Editorial pen name

Old Parker covers the Premier League, Champions League, Europe's major leagues and international football. He reads matches through schedule load, home-field context and recent form, then uses the numbers to explain formats, qualification routes and viewing times for readers in Taiwan.

View profile and articles ›

‹ Back to guides · Glossary