Right, let's get this straight: expected goals (xG) is probably the most misunderstood stat in modern football. Punters see it quoted after matches—"Liverpool had 2.3 xG but only scored one"—and think it's some kind of excuse. It's not. It's actually a window into whether a team genuinely played well or just got lucky.
Here's the thing: shots aren't created equal. A tap-in from two yards out isn't the same as a 30-yard speculative effort. Expected goals assigns a probability to every shot based on historical data—location, angle, defensive pressure, type of assist. A clear-cut chance might be 0.6 xG. A half-chance from distance might be 0.05. Add them all up across a match and you've got a team's xG total.
The beauty is this: over time, xG correlates with actual goals more reliably than raw shot count. It cuts through the noise. And for bettors, that's gold.
Expected Goals vs Shots: Why the Difference Matters
I'll give you a concrete example from the World Cup here in June. Spain played Germany in the group stage—both technically excellent teams. Spain finished with 15 shots, Germany with 11. Your instinct: Spain dominated. But Spain's xG was 1.8. Germany's was 2.1. Germany created higher-quality chances despite fewer attempts. Spain's possession masked the fact that they were wasteful.
This is why a casual glance at shot stats will cost you money.
A shot is binary: it either happened or it didn't. xG is granular. It says: "Based on where this shot came from and the circumstances, there was a 35% chance it went in." Across 10 shots of that quality, you'd expect roughly 3.5 goals. It's predictive, not descriptive.
Here's the practical angle: when you see a team with high xG but low actual goals, they're either unlucky (which regresses) or playing against a keeper having the game of his life (which also regresses). Either way, they're a value bet to improve next time. Conversely, a team with low xG and high goals? That's unsustainable. The odds will catch them.
How Expected Goals Actually Works
xG models use machine learning trained on thousands of shots from top leagues. They factor in:
- Shot location: Distance and angle from goal. A shot from 15 yards centrally is inherently more dangerous than one from 25 yards wide.
- Defensive pressure: How many defenders are nearby. A clear sight of goal gets higher value than a blocked attempt.
- Assist type: How the ball arrived. A cutback from the byline scores more often than a loose ball in midfield.
- Shot type: Header vs foot, one-touch vs controlled. Headers from close range convert at predictable rates.
- Game context: Whether it's early or late, scoreline, opposition quality. These matter less than you'd think, but they're in the model.
Different models (StatsBomb, Understat, InStat) weight these variables slightly differently, so you might see xG vary by 0.1 or 0.2 depending on source. That's normal. The direction is what counts.
The result is a number like 1.73 xG per team. That's not "expected goals scored"—it's the cumulative probability that those shots should have converted at. A 0.8 xG shot is still only a miss 20% of the time. But across a season, actual goals track xG surprisingly tightly.
Why xG Matters for Betting
Let me be blunt: if you're not using xG data, you're leaving money on the table.
The bookmakers price odds based on recent form, perceived quality, and public betting patterns. They don't always account for underlying performance. This creates gaps.
Example: Brazil played France in the World Cup knockout stage (we're in late June now, and it's still genuinely competitive). Brazil won 2-1 but France's xG was 2.4 to Brazil's 1.9. By xG, France should have won. Odds on Brazil for the next match had already tightened because they beat France. But the xG said they were the worse team and got lucky. Taking France at slightly longer odds in a rematch scenario would be smart.
This isn't luck—it's regression to the mean. Teams regress toward their true performance level. xG tells you what that level actually is.
Practical applications for betting:
- Unders: A team with inflated goal tally but lower xG tends to regress. Backing Under 2.5 goals can be value.
- Team performance bets: A team crushing xG stats but losing matches suggests they'll pick up results soon.
- Outright tournaments: Teams with strong underlying metrics (high xG, decent defence) are safer long-term picks than teams riding hot streaks.
- Handicap markets: If Team A has 30% more xG than Team B, backing them -0.5 or -1.0 at 1.80-2.10 offers better value than straight wins.
At Winotips, we build xG trends into our model specifically to catch these spots. It's one of the strongest predictive tools available.
The World Cup 2026 Context
We're mid-tournament and xG is proving its worth. Argentina, England, Spain, and Germany all have matches ahead. France got through the group stage but their xG metrics were patchy—high possession, lower-quality chances in spells. England has been creating genuine high-danger plays (xG in the 2.2+ range per match). That suggests England's likely to progress deeper than pre-tournament odds implied, especially at the lines still available for knockout rounds.
Brazil, despite their pedigree, has xG closer to 1.6-1.8 per match despite winning. That's not elite. Don't be seduced by their name.
A Quick Recap
Expected goals is a probability metric that assigns value to each shot based on location, angle, and context. It cuts through noise better than raw stats. Over time, actual goals converge toward xG. For bettors, this means spotting when odds are misaligned with true performance. A team underperforming their xG is a bet waiting to happen. One overperforming it is a fade.
Learn to read xG charts before and after matches. Cross-reference against odds. That habit alone will sharpen your picks significantly.
Responsible Gambling: Betting involves risk. 18+ only. If gambling is affecting you, call the National Gambling Helpline free on 0808 8020 133 or visit BeGambleAware.org.