If you follow football analytics or serious sports betting communities, you have almost certainly encountered the abbreviation xG. It stands for Expected Goals, and it has transformed the way analysts, coaches, and bettors evaluate football performance. This guide explains exactly what xG is, how it is calculated, why it matters for betting, and which data sources to use in 2026.
What Is xG (Expected Goals)?
Expected Goals is a statistical metric that measures the quality of a goal-scoring chance. Every shot taken in a football match is assigned an xG value between 0 and 1, representing the probability that a typical player would score from that position under those conditions. A shot with an xG of 0.35 means that, on average, 35% of attempts from that situation result in a goal. Summing all xG values for a team across a match gives the team xG total, a measure of how many goals they should have scored based on the chances they created.
How Is xG Calculated?
xG models are built using large historical datasets of thousands of shots. The variables most commonly included are:
- Distance from goal — shots from closer range have higher xG values.
- Angle to goal — central positions offer a better scoring angle than wide positions.
- Shot type — headed shots are converted at a lower rate than shots with the foot.
- Assist type — a chance created by a cross, a through ball, a cut-back, or an individual dribble each carries different probability weights.
- Defender positioning — some models factor in the position of defenders at the moment of the shot.
More advanced xG models (often called post-shot xG or PSxG) also factor in where the ball was placed within the frame of the goal, giving a more precise view of shot quality. These models are also used to evaluate goalkeeper performance.
xG vs Actual Goals: Why the Difference Matters
A team might win 2-0 despite having an xG of 0.7 while their opponents had an xG of 2.1. This is called outperforming xG and usually indicates a combination of excellent finishing, poor finishing by the opponent, or goalkeeper heroics. Over a small sample of matches, outperforming xG is plausible. Over a full season of 38 matches, xG becomes a far more reliable predictor of team strength than actual goals scored.
The practical implication for bettors is clear: a team sitting mid-table because they have been unlucky with finishing (high xG, low actual goals) is more valuable to back than a team sitting high in the table purely because their strikers have been hot (low xG, high actual goals). The latter team is likely to regress toward their mean over time.
What Is xGA (Expected Goals Against)?
xGA is the defensive counterpart of xG. It measures the quality of chances a team concedes, regardless of how many actual goals they let in. A team with a low xGA is creating very few high-quality chances for their opponents and is genuinely well-organised defensively. A team with a low actual goals conceded figure but a high xGA is likely to concede more soon, as their goalkeeper is outperforming their expected saves rate.
How to Use xG for Betting
Finding Overvalued Favourites
Bookmakers set odds largely based on league position, media narrative, and recent results. When a team is short-priced based on actual results that do not reflect their underlying xG performance, there is often value on the opposition or on Under markets. For example, a team that has won four games in a row but with an average xG of only 0.9 per game is less reliable than their odds suggest.
Targeting Over and Under Goals Markets
Pre-match xG models give a more precise prediction for the total goals line than any other method available to recreational bettors. If your model projects a combined xG of 3.4 for a match priced at Under 2.5, you have a clear signal to back Over 2.5.
Evaluating BTTS Markets
xG by team gives you a split view. If Team A is generating 1.6 xG per game and Team B is generating 1.3 xG per game, the combined picture strongly suggests both teams will score. If one team is generating only 0.6 xG per game, BTTS No becomes more attractive regardless of what the raw results have been recently.
Best Data Sources for xG in 2026
| Platform | Coverage | Cost |
|---|---|---|
| FBref.com | Top 5 leagues plus many others | Free |
| Understat.com | Top 6 European leagues | Free |
| Opta / Stats Perform | Global, professional grade | Paid |
| Sofascore | 200+ competitions | Free (basic) |
| FootyStats.org | Top 30+ leagues | Free and paid tiers |
For most recreational bettors, Understat.com and FBref.com provide more than enough data to build a solid xG-based betting approach at no cost.
Limitations of xG
- xG models differ between providers, so an xG of 0.4 on one platform may be 0.35 on another for the same shot.
- xG does not capture individual player quality, as a world-class striker will outperform xG sustainably over large samples.
- xG is backward-looking and tells you what happened, not what will happen. Combine it with qualitative factors.
- Small match samples (under 10 games) make xG figures noisy and less reliable.
Frequently Asked Questions
Is a high xG always good?
A high xG for is good, as it means the team is creating quality chances. A high xGA (against) is bad, as it means the team is allowing good chances. Always specify whether you mean xG for or xG against when discussing a team performance.
What is a good xG per game for a football team?
In the top European leagues, an xG of 1.5 per game or above is considered strong attacking output. Elite teams like Manchester City or Bayern Munich regularly generate 2.0+ xG per game over a full season. An xG below 1.0 per game suggests a team is struggling to create meaningful scoring chances.
Do bookmakers use xG to set their odds?
Yes, sophisticated bookmakers use xG-based models as part of their odds-setting process. This means that obvious xG signals are already partially priced in by the time markets open. The edge for recreational bettors comes from combining xG with contextual factors that models may not fully capture, such as squad news and tactical changes.
How many matches do I need to trust xG data?
Most analysts consider xG data reliable after approximately 10 to 15 matches for a single team. Below that sample size, random variance in finishing and goalkeeping can distort the picture significantly. Always check the sample size when reviewing xG statistics, especially early in a season.
Can I use xG for leagues outside Europe?
Yes, though data availability varies. The MLS, Brazilian Serie A, and Argentine Primera Division are covered by several platforms. Coverage for lower-tier or Asian leagues is more limited. For competitions with limited xG data, rely on goals-based statistics and qualitative scouting instead.
