xG, xA and xT: What These Numbers Actually Tell Us
Expected goals is not a prediction. It measures chance quality — and here is where it misleads.
Expected goals (xG) is a number between 0 and 1 estimating the probability that a given shot becomes a goal, based on location, shot type, pressure and body part. A value of 0.25 means one in four similar shots score.
What it measures well
xG measures process rather than outcome. A team creating 2.4 xG and losing 0-1 usually played better than the score suggests. A team winning 3-0 with 0.8 xG is often relying on elite finishing or a poor goalkeeping performance.
What it measures badly
The model does not know who is shooting. It does not know whether the chance fell to an elite striker or a centre-back. It ignores pitch condition, weather and the psychology of the 93rd minute.
The most common error is using xG for a single match. Over 90 minutes randomness dominates. The signal appears over 10 to 15 games.
xA and xT
Expected assists measures the probability a pass becomes an assist. Expected threat is broader: how much an action increases the team’s scoring probability, even if it is not the final pass.
A midfielder who breaks a pressing line with a 25-metre pass may never register an assist, yet his xT per game is high. That is why these metrics are better for evaluating central players than wingers.
Using them properly
Compare xG with actual goals across a season, use xG per shot to judge chance quality, and always add context: opponent, venue and stakes. Statistics do not replace watching — they help you see what emotion hides.