Expected goals, explained without the maths

Few statistics have been argued about as loudly, or understood as poorly, as expected goals. Most of the heat comes from people arguing about something xG does not claim to measure.

What the number means

Every shot is compared against a large historical database of similar shots and assigned the share of those that were scored. The inputs typically include:

• Distance from goal
• Angle to goal
• Body part used (foot, head, other)
• Type of assist (through ball, cross, cutback, set piece, rebound)
• Defensive pressure and number of defenders between ball and goal
• Whether it followed a dribble past a defender

A shot worth 0.15 xG means shots from that situation are scored roughly 15% of the time. Nothing about the specific striker is in the number — and that is a feature, not an oversight. The point is to describe the chance, independent of who happened to be taking it.

Rough reference points

Situation	Approximate xG
Penalty	~0.76
Close range, central, unpressured	0.35–0.50
Header from a cross, six-yard box	0.15–0.25
Edge of the box, central	0.05–0.08
Long range or tight angle	Below 0.03

The penalty figure is the most useful thing on this table, because it gives you a scale to calibrate against. A chance has to be genuinely extraordinary to be worth as much as half a penalty — and almost nothing in open play reaches 0.5.

Reading a match total

A rough guide to what a team’s ninety-minute xG total means:

Match xG	Interpretation
Under 0.7	Created essentially nothing
0.7–1.2	Below average attacking output
1.2–1.8	A normal, competitive performance
Over 2.5	Dominant; losing from here is genuinely unlucky

What it is good for

Over a season, xG predicts future scoring better than actual goals do, because it is less distorted by a handful of unrepeatable finishes. A team consistently out-performing its xG is usually about to stop; a team under-performing it while creating the same chances is usually about to improve.

It is also the fastest way to see whether a side is genuinely creating or merely shooting. Twenty shots worth 0.8 xG in total is not a dominant performance — it is a team taking bad shots from distance because it cannot break the defence down. The shot count says one thing; the xG says the truth.

What it cannot tell you

• Anything reliable about a single match. Variance dominates over ninety minutes. A 2.1 to 0.4 xG defeat is a completely normal event, not an injustice.
• Finishing quality. It deliberately excludes the identity of the shooter, so it systematically undervalues players whose finishing genuinely is exceptional.
• The pass before the shot. A model sees the chance, not the vision that produced it.
• Goalkeeper quality. Which is why post-shot xG exists as a separate metric — it measures where the shot actually went, not just where it was taken from.
• Game state. A team 2–0 up creates fewer chances on purpose. Its xG will look poor and its management will look excellent.

The most common misuse

Quoting a single match’s xG as proof that the result was wrong. It is not. xG describes the quality of chances created; the result describes what happened. Both are facts, and the gap between them over ninety minutes is normal rather than scandalous.

Over thirty-eight matches, that gap closes. That is the only timescale on which xG is making a claim.

How to use it well

Treat it as one input among several. A team with high xG and poor results has a finishing problem or bad luck, and the difference matters. A team with low xG and good results is riding something that will not last. A player consistently over-performing his xG for several seasons across different teams probably is a genuinely elite finisher — which is exactly the conclusion xG is best placed to help you reach, and only over that timescale.