A striker runs toward goal, receives a cross from his winger, takes a great first touch, and dribbles past the goalkeeper. Just before shooting, however, he missteps and the ball goes wide: no goal. Later in the match, a midfielder takes a gamble and shoots from 30 meters out, and the ball goes in. Which chance was “better”? In modern football analytics, this question is answered with a single number: expected goals, or xG. Behind this widely used but seemingly simple metric lies a classic econometric model. But how is this probability actually estimated?