Expected goals (xG) is a statistical model that assigns each shot a probability of resulting in a goal, based on shot location, body part, angle, and whether it was assisted. A shot worth 0.30 xG has a 30% chance of going in based on historical data for equivalent shots. When a team's xG significantly exceeds the actual goals scored, they were unlucky — and their next results are likely to mean-revert. Bettors use xG to identify teams the market systematically misprice due to scoreline bias.
What Is Expected Goals (xG)?
xG is assigned per shot using a model trained on thousands of historical shots. Each shot location, shot type (header vs foot), angle to goal, and shot context (open play, set piece, counter-attack) contributes to the probability estimate. A penalty has approximately 0.76 xG; a header from 18 yards at a tight angle might have 0.04 xG.
Match xG is the sum of all shot xG values across the 90 minutes. A team that created shots totalling 2.4 xG but scored 1 goal was unlucky relative to their attacking performance. A team that scored 3 goals from 0.8 xG was extremely fortunate.
Multiple providers publish xG data: Understat (free, European leagues), FBref (free, StatsBomb-powered), StatsBomb (professional tier), Opta (media/commercial).
xG vs Scoreline: The Bettor's Use Case
The betting market prices teams primarily from results, not underlying performance. A team that loses 0–2 but generates 2.1 xG to their opponent's 0.4 xG is not a bad team — they were unlucky. The market will often price the next match too long for the "loser" and too short for the "winner."
A consistent xG outperformance pattern over 5–10 matches provides evidence of:
- Poor finishing (actual goals much lower than xG) — team likely to regress positively
- Shot-stopping variance (xG against much higher than goals conceded) — goalkeeper running hot, likely to revert
- Structural quality that the market undervalues due to results focus
Applying xG to Betting Decisions
xG Rank vs League Rank
Compare a team's xG rank to their actual points rank. A team ranked 12th in league position but 5th in xG differential is outperforming on the table — they are expected to climb. A team ranked 4th in points but 14th in xG has been outperforming luck; they are a regression candidate.
This gap between underlying quality and market-facing position is where value frequently exists in mid-season match betting.
Recent Form vs Recent xG
Public perception (and many bookmakers' power ratings) uses the last 5 results heavily. Compare the last 5 results to the last 5 xG differentials. A team on a 3-loss run but with positive xG in each match is more likely to win their next than the odds suggest.
Head-to-Head xG History
Some matchups produce structurally low-xG games (defensive teams, low-tempo tactics) or high-xG games (open play, frequent chances). If you are betting totals (over/under goals), historic xG averages for the specific matchup are more predictive than league averages.
Limitations of xG Betting
xG is a useful tool, not an oracle. Key limitations:
- Individual striker quality — elite forwards systematically outperform xG; poor finishers underperform. xG is calibrated to average forwards. Erling Haaland outperforms his xG because his shot quality genuinely exceeds the model's average.
- Goalkeeper quality — elite keepers save shots the model predicts as goals. Over a full season, the best goalkeepers consistently underperform xGA.
- Model variation — StatsBomb, Opta, and Understat use different feature sets. The same shot may receive different xG values across models.
- Information lag — if a quality xG model is publicly available and free, efficient markets will incorporate it into odds pricing. The edge is largest in lower leagues and early in the season before the market updates.
Where to Find xG Data
- Understat.com — free, match-by-match and player xG for the top 6 European leagues
- FBref.com — StatsBomb-powered, free public data for most major global leagues
- SofaScore / WhoScored — include xG in match stats, sourced from Opta
- 365Scores — mobile-friendly xG in live stats
FAQ
- Is xG reliable for betting?
- xG is a useful predictive signal, not a guaranteed edge. Markets in the top leagues increasingly price xG into odds, reducing the edge. The strongest xG-based edges tend to exist in early-season pricing (before sample sizes accumulate), in lower-division markets (where bookmakers rely on simple models), and in total goals markets where shot volume and quality data is underused.
- What is a good xG differential?
- A positive xG differential (xG for minus xG against) above +0.5 per match over 10+ games indicates a team generating significantly more chance quality than they concede. The best teams (Man City, Arsenal, Liverpool) typically run +0.8 to +1.2 xG differential per match over a full season. Midtable is typically +0.1 to -0.1. Below -0.4 per match is relegation territory.
- Does xG account for penalties?
- Most xG models assign a fixed value to penalties — typically 0.75–0.79. This means penalty-heavy teams accumulate xG for penalties, which is valid: they earn the right to convert. However, penalties are rarer than open play shots, so a season's xG from penalties versus open play can indicate whether a high-scoring team is genuinely dominant or penalty-dependent.
- What does negative xG mean for my bet?
- Negative xG differential means a team is conceding better chances than they create. As a bettor, this indicates a team that relies on individual brilliance or goalkeeper performance to win — neither is sustainable. Teams with consistent negative xG differentials in their wins are good short candidates once the luck evens out.
- Do bookmakers use xG in their pricing?
- Major bookmakers and trading firms use versions of xG as inputs to their pricing models, especially for in-play markets and next-match odds. The degree to which this is fully priced in varies by market and league. Lower leagues and international competitions tend to have larger xG-based edges because bookmakers devote less modelling resource to them.