Expected Points

Contents

What is expected points?

Expected points (xP) is a way to measure how many points a team should have earned based on the chances they created and gave away during a match, not just the final score.

To do this, it uses something called expected goals (xG). This looks at every shot and gives it a score based on how likely it was to result in a goal. It considers things like how far the shot was from goal, the angle of the shot and what type of shot it was (like a header or volley)

Using this xG data, xP runs lots of match simulations to figure out what the likely result should have been. This gives a better idea of how well a team actually performed, rather than just looking at wins or losses.

How are expected points (xP) calculated?

Expected points (xP) are worked out using a step-by-step process based on expected goals (xG) and computer simulations of the match.

Here’s how it works:

1. Collecting xG data

Every shot in the match gets a score called xG, which shows how likely it is to become a goal. This score is based on things like how far the shot is from goal, the angle of the shot, the type of shot (like a header or volley) and what’s happening in the game (for example, pressure or the scoreline)

2. Running match simulations

Computers use this xG data to simulate the match thousands of times. Each time, random numbers decide whether a shot goes in or not based on its xG score. This gives a spread of likely results: win, draw, or loss.

3. Calculating the expected points

From the simulations, we get the chance of each result (like 50% chance of winning, 30% chance of drawing).

These are plugged into a simple formula:

xP = 3 × chance of win + 1 × chance of draw (Losses count as 0)

For example: If a team has a 50% chance of winning and a 30% chance of drawing: xP = 3 × 0.5 + 1 × 0.3 = 1.8 points

4. Adjusting for multiple shots in the same attack

If a team takes several shots during the same move, we don’t count them all as separate chances. Instead, the model works out the combined chance of scoring in that possession.

For example, if there are 3 shots with xG scores of 0.2, 0.4, and 0.3, the chance of no goal is: (1 – 0.2) × (1 – 0.4) × (1 – 0.3) Then we subtract that from 1 to find the chance of a goal.

This makes sure the numbers stay realistic and not inflated.

Why expected points (xP) matter

Expected points (xP) help us understand how well a team is really playing, not just how many points they’ve collected. It’s a powerful tool in football analysis for several reasons:

1. Spotting over and under-performance

xP shows when teams are getting lucky or unlucky.

– For example, in the 2024/25 season, Nottingham Forest earned around 15 more points than their xP predicted (65 points vs 50 xP), which means they might’ve been a bit lucky or very efficient.
– On the other hand, Manchester United got 10 fewer points than their XP suggested, meaning they may have been unlucky or wasteful.

This helps show the difference between results and real performance.

2. Predicting future results

When a team performs much better or worse than their xP, it often doesn’t last forever. Teams usually return to their “true” level over time. This is called regression to the mean.

So, xP can help predict if a team will improve or decline in the future and inform long-term betting decisions

3. Judging tactics and coaching

xP tells us if a coach’s playing style is creating good quality chances, even if the team isn’t winning. This helps coaches evaluate if their tactics are working and make better decisions to improve team performance

4. Helpful in betting and scouting

xP can uncover teams that are likely to do better (or worse) than current results show.

5. For scouts and recruiters

xP helps find players or teams performing well in a way that’s likely to last, not just short-term form.

Real-world examples

Expected Points (xP) has been applied across various football leagues to highlight discrepancies between actual and expected performance, providing valuable insights. Notable examples include:

– Bournemouth (2024/25): Bournemouth’s XP significantly exceeded their actual points (9 points difference), suggesting they were unlucky in converting high-quality chances into goals. If the EPL table were based off of XP, they would have comfortably made Champions league football this season.
– Wolverhampton Wanderers (2022/23): Wolves would have relegated to the English Championship if the xP table was used as they gathered only 35 points on xP, against the 41 points they did in the league. This indicated that they got lucky with finishing and were often second best on multiple occasions.
– Manchester City (2023/24): xP data shows that City were the right heirs to the EPL trophy last season (look away Arsenal fans 👀). They did get 8 less points, according to xP data, but so did Arsenal, who got 7 less points. The final standing would have been City, 83, Arsenal, 81.
– Liverpool (2019/20): We have to go all the way back to the 2019 season to see a big upset. It was the season after City had just done a league double. Liverpool went on to win the league by 18 points, finishing with 99 points, 1 shy of City’s centurions the year before. XP data however tells a different story. It shows that Liverpool has an xP of just 75 points, and City (poor by their standards that season, has a better xP at ~ 87 points.

Limitations and critiques of expected points (xP)

Even though expected points (xP) is a useful tool, it’s not perfect. There are some important things to keep in mind when using it:

1. It depends on xG being accurate

xP is based on expected goals (xG), so if the xG model isn’t accurate, the xP won’t be either.

For example, if the model overrates or underrates certain types of shots (like long-range shots or headers), it can give a false picture of how well a team really played.

2. It can be unreliable in small samples

xP works best over longer periods. If you’re looking at just one match or a short stretch of games, it can be misleading. A team might finish well or poorly in a few games, which can skew the numbers. The model might not reflect true performance in the short term.

3. It doesn’t always include full match context

Most xP models don’t fully consider things like how a team keeps possession, how their tactics change during the game, what happens when a team is winning or losing (like sitting back after taking the lead). This means xP can sometimes oversimplify what really happened on the pitch.

4. It doesn’t fully account for randomness

Football has a lot of unpredictable moments, lucky bounces, bad refereeing calls, or one-off mistakes. xP can’t always capture that.

It also might miss how teams change their approach mid-game, which affects the types and quality of chances they create or allow.

Advanced models and where xP is heading

Expected points (xP) models are getting better all the time. New methods are helping to fix earlier problems and give even deeper insights. Here’s what’s changing, and what the future may look like:

1. Using more than just xG

Older models focused only on expected goals (xG). Newer ones now also include:

– Expected threat (xT): This tracks how dangerous non-shooting actions are, like passes or dribbles into key areas.
– Possession-based stats: tracking how a team builds up play and controls the game, even without shooting.

This gives a fuller picture of performance.

2. Pythagorean-style models

Some models are now using ideas from baseball’s Pythagorean expectation.

– These models use season-long data like goals scored, xG, and xGA (expected goals against).
– They try to predict long-term success rather than focusing on single matches.
– Since football has more draws than baseball, these models adjust for that.

It’s a more big-picture approach to understanding performance.

3. Tracking players and tactics

Future models are starting to use detailed tracking data, such as where players move, how teams change shape and how game situations affect decisions

This makes xP models smarter and more accurate, capturing the full context of each match.

4. Making models easier to understand

It’s not just about being clever, it’s also about being clear.

Analysts and coaches want models they can trust and explain. That’s why future XP tools are being designed to be more transparent and easier to use, especially for real-world decisions like choosing tactics, scouting players and making recruitment decisions.

Sportmonks and expected points (xPTS)

At Sportmonks, we enhance our football API with robust expected points (xPTS) data, alongside the more familiar expected goals (xG) and related metrics. Their documentation defines xPTS as:

Access & endpoints

GET /football/expected/fixtures (by team) retrieves xPTS along with a range of expected values (xG, xGA, xGFK, xGP, etc.) at fixture level.

Standings endpoint with details.type include, returns season-level xPTS rankings integrated into conventional standings data.

Coverage & tiers

xPTS is part of the full range of expected values offered alongside: xG, xGoT, xGA, xGP (penalties), xGFK, xGC (corners), xGSP (set plays), npxG, and others.

API plans (Basic, Standard, Advanced) control timing:

– Basic: data available ~12 hours after match end.
– Standard: immediate post-match.
– Advanced: real-time, including live xPTS values.

Use cases with Sportmonks xPTS

– Fixture analysis: Use match-level xPTS to contextualise results beyond simple win/draw/loss, identifying teams that “should have” earned more or fewer points.Season insights: The enhanced standings include xPTS for every team, enabling detection of overperformers and underperformers, highlighting squads outperforming their underlying quality or ones riding luck.
– Easy integration: A single API request with GET fixtures?include=xGFixture or GET standings?include=details.type embeds xPTS into existing data structures.

GET https://api.sportmonks.com/v3/football/standings/seasons/{season_id}?include=details.type

This returns lines in the standings enriched with type_id: 7939 (xPTS) and the expected points value, perfect for side‑by‑side comparison with actual points.

Why include Sportmonks xPTS in your analysis

– Unified metric suite: xPTS is packaged together with xG and other expected metrics, simplifying data pipelines and enriching analytical depth.
– Flexible access: Whether analysing individual matches (fixture-level) or full seasons (standings-level), Sportmonks provides xPTS seamlessly.
– Real-time value: With advanced subscription tiers, live xPTS is available, unlocking near-instant tactical insights and reactive analysis.

Rethink the table: Go beyond goals with Sportmonks’ xPTS data

Expected points (xPTS) tell the real story behind a team’s performance, beyond what the final score says. With Sportmonks’ football API, you can plug live and post-match xPTS data directly into your analysis tools, betting models, or content platforms.

Whether you’re tracking overachievers or spotting unlucky underdogs, Sportmonks offers detailed xPTS data at fixture and season level. It pairs seamlessly with metrics like xG, xGA, and xGoT to paint a deeper picture of team quality.

Add smarter insights to your app or dashboard

Sportmonks makes it easy to integrate xPTS into your site or app. From visualising luck-adjusted tables to breaking down performance trends across leagues, you’ll help users see what the scoreboard misses.

Start exploring expected points with the Sportmonks football API today and give your audience the full story behind every result.

How are expected points calculated?

Expected Points (xP) are calculated by taking the Expected Goal (xG) values for each shot in a match and converting them into a probability of winning, drawing, or losing. This is typically done through a simulation process. For every shot taken by a team, the xG value (probability of that shot being a goal) is used to determine the likelihood of scoring a certain number of goals. These probabilities are then aggregated for both teams to determine the most probable final scorelines, and each scoreline is assigned a point value (3 for a win, 1 for a draw, 0 for a loss). The sum of these point values multiplied by their probabilities gives the expected points for a team in a given match. Essentially, it estimates how many points a team should have earned based on the quality of chances they created and conceded.

Is xG a good predictor?

Yes, Expected Goals (xG) is generally considered a very good predictor, especially over a larger sample size of games. While it doesn't perfectly predict the outcome of a single match (due to luck, goalkeeping heroics, or individual brilliance), xG is a strong indicator of underlying team performance. Over a season, teams with consistently higher xG differences (xG for minus xG against) tend to climb the league table, as it reflects their ability to create good scoring opportunities and limit their opponents'. It's a more stable and reliable measure of offensive and defensive quality than raw goals scored or conceded, which can be heavily influenced by variance.

What does "expected points added" mean?

"Expected Points Added" (xPA) is a concept more commonly found in American sports analytics (like football or basketball) and refers to the change in a team's expected probability of winning, drawing, or losing based on a specific play or event. In soccer, while the term "xPA" isn't as widely used as xG or xP, the underlying idea can be applied. For example, a high-quality shot (high xG) would "add" to a team's expected points for that match, as it significantly increases their chances of scoring and thus winning or drawing. Conversely, a defensive error leading to a clear-cut chance for the opponent would "subtract" from a team's expected points. It quantifies the impact of individual actions on a team's overall expected outcome.

How do you calculate expected goals?

Expected Goals (xG) are calculated using historical data of thousands of shots and a variety of factors related to the shot's context. A machine learning model (often a logistic regression or similar) is trained on this data to assign a probability of a shot resulting in a goal. Key factors that go into calculating xG include: - Shot location: Distance from goal and angle to goal are crucial. - Body part: Headed shots typically have lower xG than shots with feet. - Type of assist: Through balls, crosses, cutbacks, and rebounds all influence xG differently. - Defensive pressure: Number of defenders between the shooter and the goal, and how close they are. - Game state: Open play, set piece, counter-attack, etc. - Goalkeeper position: Though this is harder to model consistently. Each shot is assigned an xG value between 0 and 1, representing the likelihood (e.g., an xG of 0.5 means there's a 50% chance that shot will result in a goal based on historical data of similar shots).

Written by Wesley Van Rooij

Wesley van Rooij is a marketing and football expert with over 5 years of industry experience. His comprehensive knowledge of the Sportmonks Football API and a focused approach to APIs in the Sports Data industry allow him to offer insights and support to enthusiasts and businesses. His outstanding marketing and communication skills and technical writing expertise enable him to empathise with developers. He understands their needs and challenges to facilitate the development of cutting-edge football applications that stand out in the market.