Key takeaways
- xG, or expected goals, measures the probability that a shot will result in a goal, based on historical data from thousands of similar attempts.
- A shot has an xG value between 0 and 1: a penalty is roughly 0.76 xG; a header from 30 yards might be 0.01 xG.
- xG helps separate luck from quality, showing whether a team's goal tally reflects their actual performance or just fortunate finishing.
- Related metrics include xGA (expected goals against), npxG (non-penalty expected goals) and xA (expected assists).
- xG is now used by clubs, broadcasters and fantasy football managers to evaluate players and predict future results.
---
xG, or expected goals, is a statistical measure that assigns every shot a probability score between 0 and 1 reflecting how likely it is to become a goal. A score of 0.1 means a 10% chance; a score of 0.9 means a 90% chance. The values are calculated using factors like shot location, assist type and whether the chance was a header or a foot shot, all calibrated against large historical datasets.
As of June 2026: what's current
xG is now a mainstream metric across European football. BBC Sport and Sky Sports include it in match broadcasts, and most top clubs employ analysts who build on or extend the public xG models available through sites like FBref and Understat. The metric has become a standard part of Champions League coverage and fantasy football strategy alike.
---
What exactly is xG and where did it come from?
xG, or expected goals, is a football analytics metric that assigns a numerical probability to each shot attempt, telling you how likely that specific chance was to be scored, based on historical data from thousands of comparable shots.
The concept was developed in the early 2010s. Analyst Sam Green presented an early version at the 2012 MIT Sloan Sports Analytics Conference, and the model was later developed further by companies including Opta and StatsBomb. By the mid-2010s, clubs in the Premier League and across Europe were using proprietary versions internally. Public xG data became widely available around 2014 to 2016, when sites like Understat began publishing match-level figures.
The core idea is simple: not all shots are equal. A tap-in from two yards is almost certain to go in; a speculative attempt from the halfway line almost never does. xG puts numbers on that difference.
---
How is xG actually calculated?
xG models are built by feeding historical shot data into a statistical model, typically a logistic regression or a machine learning algorithm, and finding the factors that best predict whether a shot becomes a goal.
The main variables most models use include:
- Distance from goal. Shots from inside the six-yard box score much higher than those from outside the box.
- Angle. A central position creates a better chance than a tight angle.
- Shot type. Headed shots are converted at a lower rate than shots with the foot, so they receive lower xG values.
- Assist type. A cut-back pass or a through-ball tends to produce higher-quality chances than a long-range cross.
- Situation. Open play, set pieces and penalties are all modelled separately.
Penalties are worth roughly 0.76 xG under most public models, reflecting the historical conversion rate of around 75 to 78 percent. A big chance, defined by Opta as a situation where the player should reasonably be expected to score, typically carries an xG value above 0.35.
FBref, which uses StatsBomb data, is one of the most detailed public sources. Understat covers the top five European leagues with match-by-match xG breakdowns going back several seasons.
---
What do the related metrics mean?
xG sits at the centre of a family of metrics. Here is a quick reference table.
| Metric | Full name | What it measures |
|---|---|---|
| xG | Expected goals | Probability a shot results in a goal |
| xGA | Expected goals against | xG from shots the opposition has taken |
| npxG | Non-penalty expected goals | xG with penalties removed |
| xA | Expected assists | Probability a pass leads to a goal |
| xGD | Expected goal difference | xG minus xGA for a team |
| PSxG | Post-shot expected goals | xG recalculated after watching where the shot goes |
| xGOT | Expected goals on target | Expected value of shots that hit the target only |
npxG is often preferred when comparing forwards because penalties inflate raw xG totals. A striker who takes six penalties in a season looks more productive in raw xG terms, but npxG strips that out and gives a cleaner read of their open-play and set-piece threat.
PSxG (post-shot expected goals) is used to evaluate goalkeepers. It measures how dangerous shots were by the time they were struck, accounting for placement and power, not just the original position. A goalkeeper who allows fewer goals than the PSxG they faced is performing above expectation. FBref publishes PSxG data for every goalkeeper in the major leagues.
xA, or expected assists, is a metric that assesses the quality of a pass leading to a shot, not just whether it generated an assist. A player who creates three big chances but none are converted will have a high xA but zero actual assists. That gap between xA and actual assists often signals a player whose output is being undervalued.
---
How should you read xG in a match context?
The most useful application is comparing a team's xG over a full season, not just one game. Over 90 minutes, a team can outscore their xG through excellent finishing or by conceding goals from very low-quality chances. Over 38 league games, those fluctuations even out.
A team with a strong positive xGD, meaning they consistently create more quality chances than they concede, tends to finish higher in the table over time. According to analysis published by The Guardian, xG difference is a stronger predictor of future league position than actual goal difference at the midpoint of a season.
Some practical reading guidelines:
- A match xG of 2.5 vs 0.8 suggests the winning team genuinely dominated, regardless of the scoreline.
- A match xG of 1.0 vs 0.9 was essentially an even game; the result could easily have gone either way.
- A striker finishing at 150% of their xG over half a season is likely benefiting from above-average finishing form that will regress toward the mean.
Sofascore and FotMob show live and post-match xG for most major leagues, making it easy to check a game's underlying quality in real time.
For deeper tournament previews that already have xG baked in, our Champions League 2026/27 power rankings and favourites uses expected goals data to assess squad quality across Europe's top clubs.
---
Which teams and players benefit most from xG analysis?
Teams that outperform their xG
Some clubs consistently score more goals than their xG suggests. This can reflect genuinely elite finishing in the squad, but it more often signals unsustainable luck. When analysts see a club sitting 10 points clear of their xGD-implied position in November, they tend to expect a correction.
Why they matter: Identifying overperforming teams early helps predict second-half-of-season collapses and informs transfer strategy. A club buying a striker because they scored 18 goals may be overpaying if that player only generated 11 xG.
Key stat: Across the top five European leagues, teams that outperformed their xGD by more than 0.4 goals per game in the first half of a season have historically regressed toward their expected position in the second half around 70% of the time, according to long-run analysis on FBref.
Forwards with high npxG
Midfielders with elite xA
For midfielders, xA reveals creative output that standard assist counts bury. A player who creates 12 big chances but plays in a team with poor finishing will show near zero assists but a very high xA. That is exactly the kind of undervalued profile that analytics-led clubs target in the transfer market.
Best for: Identifying playmakers who are creating value that doesn't show up in traditional stats.
Key stat: StatsBomb data shows that xA above 0.20 per 90 minutes puts a midfielder in roughly the top 5% of central midfielders across the Big Five leagues.
Goalkeepers who save above PSxG
Goalkeepers who consistently prevent more goals than their PSxG predicts are performing above the average goalkeeper facing the same shots. Over a full season, that can amount to five to eight goals prevented, which is a significant competitive advantage.
Best for: Separating genuinely elite shot-stopping from goalkeepers who face few dangerous attempts because their defence is strong.
---
xG in fantasy football and betting: how useful is it really?
xG has become a core tool for Fantasy Premier League managers and for serious football bettors. Its main value in both contexts is the same: it strips out noise and gives you a more stable picture of underlying performance.
For FPL, a forward who has scored two goals from 4.5 xG over the last four gameweeks is in a much stronger position than their blank returns suggest. Their chances have been good; the ball hasn't gone in. Holding or buying them before a regression is a legitimate strategy.
For pre-match analysis, xG-based models produce implied probabilities for match outcomes that sometimes differ from bookmaker odds, particularly in matches involving smaller clubs or cup competitions where bookmakers have less pricing efficiency.
Our guide to the best football prediction sites and models compares tools that use xG and other advanced metrics for match forecasting.
It is worth being direct about the limits here. xG models do not capture everything. Tactical disruptions, key absences, weather, and morale can all shift a match result in ways that xG will not pick up. It is a probability tool, not a certainty engine.
---
A comparison of the major public xG data sources
Different providers build their models differently, and the figures they publish for the same match can vary. Here is a comparison of the main public sources.
| Source | Leagues covered | Model type | Free access | Goalkeeper PSxG | Player-level data |
|---|---|---|---|---|---|
| FBref (StatsBomb) | 30+ competitions | StatsBomb proprietary | Yes (basic) | Yes | Yes |
| Understat | Big 5 leagues | Gradient boosting | Yes | No | Yes |
| Sofascore | 100+ leagues | Proprietary | Yes | Partial | Yes |
| FotMob | 100+ leagues | Proprietary | Yes | No | Yes |
| Opta / WhoScored | Big 5 + more | Opta proprietary | Partial | Yes | Partial |
For most users, FBref provides the deepest free dataset for player and team-level xG, while Understat is the quickest way to check match-level figures for the top European leagues. Our full breakdown of these tools is in the best football stats sites and apps guide.
---
Frequently asked questions
What is a good xG in football?
An xG of 1.0 per game is roughly average for a top-flight club. A figure above 1.5 per game over a full season suggests a team is generating high-quality chances consistently. For individual players, an xG of 0.30 per 90 minutes is broadly considered a strong return for a forward.
Can xG be negative?
No. xG is a probability score and always sits between 0 and 1 per shot. A team's cumulative xG for a match can be lower than their opponent's, which sometimes gets described informally as "negative" xG performance, but individual xG values are always positive.
Why does xG sometimes look wrong after a match?
xG reflects the average outcome from similar historical shots, not the specific shot on the day. A goalkeeper can make an extraordinary save on a 0.8 xG chance. Over many matches those deviations average out, but single-game xG figures can feel misleading if you focus only on the final score.
Is xG used by professional clubs?
Yes. According to reporting by ESPN, the majority of Premier League and top European clubs employ analysts who use xG and related metrics as part of scouting, recruitment and tactical preparation. Most clubs use proprietary models rather than public ones.
What is the difference between xG and xGOT?
xG is assigned at the moment of the shot based on shot location and type. xGOT (expected goals on target) is only calculated for shots that required a save, and it accounts for where the ball was heading within the goal. It is mainly used to evaluate goalkeeper performance more precisely than raw save percentage.
How does xG handle penalties?
Penalties are typically assigned a fixed xG value of around 0.76 in most public models, reflecting historical conversion rates of roughly 75 to 78 percent. Some analysts prefer to exclude penalties entirely and use non-penalty expected goals (npxG) for cleaner player comparisons.
---
The bottom line
xG is the single most useful starting point for understanding whether a football result reflected genuine performance or just variance. It will not tell you everything, but a team that consistently generates an xGD above zero is building the right foundations, and a striker finishing well below their xG over 20 games is probably due a run of goals.
If you want to put these metrics to work immediately, use the [Footballens MatchBrief tool](/app/brief) for xG breakdowns and performance summaries on any upcoming fixture. For a wider set of data tools across leagues and competitions, the best football stats sites and apps guide covers everything from free public models to professional-grade platforms.
The clubs and analysts who built competitive advantages using xG in 2015 did so because everyone else was still counting goals. The metric is mainstream now, but most casual fans still underuse it. That gap is still worth closing.
---
By the Footballens desk. Senior football writers covering the World Cup, transfers and analytics. Last reviewed June 2026.