Football Analytics

What is xG (Expected Goals)?

The metric that changed how we understand football. Learn why the scoreline doesn't always tell the full story.

8 minute read

The Problem with Scorelines

Imagine watching a match where Team A dominates possession, creates chance after chance, hits the woodwork twice, and forces the goalkeeper into a dozen saves. Team B barely crosses the halfway line but scores from their only shot. Final score: 0-1 to Team B.

The scoreline says Team B deserved to win. But did they really? This is the fundamental problem that Expected Goals (xG) solves.

xG measures the quality of chances created, not just whether they resulted in goals. It tells us what should have happened based on the opportunities each team had.

xG Explained Simply

Expected Goals (xG) assigns a probability to every shot based on how likely it is to result in a goal. The value ranges from 0 to 1:

0.02 xG

Long-range shot

2% chance of scoring

0.15 xG

Shot from edge of box

15% chance of scoring

0.76 xG

One-on-one with keeper

76% chance of scoring

A team's total xG is the sum of all their shots' individual xG values. If a team creates chances worth 2.5 xG in a match, they "should" score around 2-3 goals — though individual matches vary due to finishing quality and luck.

Think of it like this:

If you replayed the same match 100 times with the same chances, a shot with 0.3 xG would go in roughly 30 times. Some days the striker is clinical and scores from 0.1 xG chances; other days they miss sitters. xG captures the underlying quality regardless of the actual outcome.

How xG is Calculated

xG models analyze millions of historical shots to determine what factors influence scoring probability. The key variables include:

Distance from Goal

Shots from 6 yards have much higher xG than shots from 25 yards. This is the most significant factor.

Angle to Goal

Central positions have higher xG than tight angles near the byline. More goal to aim at = better chance.

Body Part

Foot shots generally have higher conversion rates than headers. Strong foot vs weak foot also matters.

Type of Assist

Through balls and cutbacks create better chances than crosses. The build-up affects shot quality.

Advanced models also consider defenders' positions, goalkeeper location, whether it's open play or a set piece, and even game state (leading vs trailing).

Real Match Example

Let's look at how xG reveals the true story of a match:

Premier League Example

Liverpool

1 - 1

Burnley

Final Score

Expected Goals (xG)

3.2

0.4

What the scoreline says

A fair draw. Both teams scored once, both deserved a point. Evenly matched contest.

What xG reveals

Liverpool dominated. They created chances worth 3.2 goals while Burnley barely threatened (0.4 xG). Liverpool were unlucky; Burnley got fortunate.

The insight: If these teams played 10 times with similar performances, Liverpool would likely win 7-8 of them. The 1-1 draw was an outlier result. xG helps us understand when a result doesn't reflect the underlying performance.

Why xG Matters

Predicts Future Performance

Teams that consistently create high xG tend to win more over a season, even if short-term results don't reflect it. xG is more predictive than actual goals for future performance.

Identifies Over/Underperformance

A striker scoring 15 goals from 8 xG is overperforming and likely to regress. A team with good xG but bad results is probably due some luck. This helps separate sustainable performance from variance.

Evaluates Teams More Accurately

A team sitting 5th with strong underlying xG numbers might be better positioned than a team in 3rd who's been clinical but creating fewer chances. xG helps identify who's genuinely performing well.

Common Misconceptions

"xG says who should have won"

Not quite. xG measures chance quality, but football rewards actual goals. The team that scores more wins — xG just tells us if that result was expected or surprising.

"xG is perfect and infallible"

xG models have limitations. They don't capture everything — the quality of the striker, defensive pressure timing, or psychological factors. It's a useful tool, not an oracle.

"One match xG is meaningful"

Single match xG has high variance. A team might have 3.0 xG from 3 headers in the box or from 30 long shots. Trends over multiple matches are more informative than single game numbers.

How We Use xG at ScoresAhead

xG is one of the foundational inputs to our prediction model. We use historical xG data to understand:

•How many quality chances a team typically creates at home vs away
•How well a team's defense limits opponents' xG
•Whether a team is over or underperforming their underlying numbers
•xG trends — is a team improving or declining in chance creation?

This historical xG data feeds into our machine learning model, which then generates pG (predicted goals) — our forward-looking prediction of how many goals each team will score. Think of xG as the foundation that makes accurate pG predictions possible.

Key Takeaways

xG measures chance quality, assigning a probability (0-1) to every shot based on historical data
Scorelines can be misleading — xG reveals whether a result was expected or an outlier
xG is predictive — teams with strong underlying xG tend to improve their results over time
It's a tool, not truth — xG has limitations and works best alongside other analysis