Expected goals, or xG, rates the quality of a shot. Every shot gets a value between 0 and 1: the probability that a shot like it becomes a goal. A shot worth 0.20 xG would be expected to go in about twice in every ten attempts, while a shot worth 0.76 xG would go in about 76 times in every 100. Add up a team's shots in a match and you get its xG: the goals an average finisher would be expected to score from those chances.

The scoreline tells you what happened. xG tells you how good the chances were, whether the goals followed or not. A team can win 1-0 while creating worse chances than its opponent. Over one match, luck and great goalkeeping can override the numbers, but over a run of matches xG is a steadier guide to which team is really creating good chances.

How xG Is Calculated

An xG model is not a person guessing how hard a shot looked. It is built from a large database of past shots, each labelled as a goal or not a goal. For a new shot, the model finds past shots that looked similar and asks what share of them went in. That share becomes the new shot's xG value.

Distance is the strongest single factor: the closer the shot, the higher the xG. Angle matters too, because a central shot has much more goal to aim at than a shot from near the byline. Body part matters as well, since headers are generally converted less often than shots with the foot from the same spot.

Chance type matters too. A shot after a through ball or cut-back is usually worth more than one after a hopeful cross, and a penalty is treated separately because every penalty starts from the same spot. Most models give a penalty around 0.76 xG, reflecting the historical conversion rate. Detailed models add defender and goalkeeper positions and whether the shooter was under pressure, which is one reason providers differ slightly for the same match.

A Worked Example: Adding Up Five Shots

Here is a worked example, using illustrative values for five shots. Imagine a team takes: a penalty worth 0.76 xG, a close-range central shot worth 0.42 xG, a shot from the edge of the area worth 0.18 xG, a header from a cross worth 0.09 xG, and a long-range effort worth 0.03 xG.

Adding step by step: 0.76 plus 0.42 equals 1.18, plus 0.18 equals 1.36, plus 0.09 equals 1.45, plus 0.03 equals 1.48. The total is therefore 1.48 xG from five shots, an average of about 0.30 per shot, since 1.48 divided by 5 equals 0.296. An average finisher would be expected to score about one and a half goals from those chances.

Compare that with twelve long-range shots, each worth 0.03 xG. Twelve multiplied by 0.03 equals 0.36 xG in total. That team took more than twice as many shots, but created far less danger. Counting shots alone rewards the team that shoots often over the team that shoots well.

Why the Same Match Can Show Different xG Figures

Slight differences between websites are normal. One model may include goalkeeper and defender positions while another does not, one may value penalties at 0.76 and another at 0.78, and each was trained on a different set of past shots. For a fair comparison, stick to one provider when comparing teams or tracking a team over time.

The same logic explains non-penalty xG, or npxG. Analysts often remove penalties to judge open-play creation on its own. In the example above, removing the penalty gives 1.48 minus 0.76, which equals 0.72 npxG: a clearer view of what the team created without the penalty.

What xG Does Not Tell You

xG only counts shots, so a dangerous attack that ends without a shot adds nothing. A basic model also treats every shooter as an average finisher, so it does not fully credit an exceptional striker or goalkeeper. Post-shot xG, or PSxG, is calculated after the shot, taking placement into account, and is often used to judge goalkeeping and finishing more closely. Even then, xG does not measure pressing or chances prevented before a shot was taken.

How to Read xG Without Overreacting

Ask three questions before drawing a conclusion. First, is the total built from many decent chances or one huge chance, such as a single penalty? Second, what was the game state? A team two goals up early will often sit deeper and let its opponent pile up low-value shots, so late xG against a parked defence is cheaper than early xG in an open game. Third, how many matches are you looking at? One match is noisy; ten matches of consistently creating more xG than you concede is a much stronger signal.

The Short Answer

Expected goals rates every shot by how likely a similar shot was to score in the past, from 0 to 1, and a team's match total is simply the sum of its shots' values. Read it beside the scoreline, not instead of it: the score says who won, and xG says whether the chances deserved that result. Over a run of matches, it is one of the clearest tools for seeing past lucky and unlucky results. For a deeper model-level explanation, see the guides from Hudl and SoccerAnalytics.