Wikipedia:Reference desk/Archives/Entertainment/2023 March 25

= March 25 =

xG
In association football, there's a metric called expected goals. As far as I understand, xG only increases when a shot at goal is taken. Is that correct? If yes, it leaves a lot to be desired. For example, a player may lose the ball with an unmarked teammate in scoring position. Is there a way to quantify these promising chances in which no shot had been taken or, in other words, is there a system already in place? Splićanin (talk) 21:31, 25 March 2023 (UTC)


 * The definitions in our article are all over the place. The one you are referring to is ascribed in the article to Ensum, Pollard and Taylor. The value of xG is estimated from a database recording only kicked shots, so it is meaningless to say that it "increases" when a shot is taken. Whether this is a useful metric depends on what it is used for. During a match, player A may face a choice, "should I take a shot at the goal (given my xG) or pass the ball to player B in a more favourable position (with a higher xG)?" Modelling this mathematically, the player should compare xGA with the product xPAB × xGB, in which xPAB stands for the chance that an attempted pass from A to B will succeed. In this (simplistic) model, the appropriate notion is indeed the expected success of a kicked shot. The value of xG, used here as a statistical predictor, does not depend on whether the shot is or is not taken. --Lambiam 15:05, 26 March 2023 (UTC)