xG is a statistical metric that has gradually risen to prominence in the last few years, now regularly being included in mainstream media, including football broadcasts.
But what is it? How is it calculated and what does it tell us?
Simply put, Expected Goals, or xG for short, is a measure of shot quality, and can be used to determine how well a team is playing. While you could look at possession as an indicator of how good a team are, having lots of the ball doesn’t necessarily make you more likely to win games. Similarly, one team could have loads of shots, but if they’re all from 30 yards out, they won’t count for much.
Expected Goals was designed to assign a value to every shot, indication how likely (or expected) it was to result in a goal.
For example, a shot from 12 yards out in the middle of the area will produce a goal more often than a shot from a tight angle from 25 yards out. As such, the first shot will have a higher xG than the second, and in the long run, teams who amass higher xG should in theory win more often.
Of course, there are many other factors in calculating xG and drawing conclusions from it. Say Team A scores an early goal and decides to sit on their lead. Team B may spend much of the game attacking and creating chances, while Team A focuses on defending. Team A may then break away and score a goal on the counter attack. While this may lead to a low overall xG, Team A may have executed their game plan perfectly. Similarly, if Team B often falls behind early in games, they’ll likely accrue higher xG than if they were leading for the majority of games.
While the metric has its limitations, xG can be used on a number of levels. While in the short term it isn’t as useful as the long term, it can offer insight into a player and team’s ability to score goals.
For example, a player who consistently over-performs their xG (scores more goals than their xG suggests they should) would typically be deemed a good finisher. If, however, a player who consistently over-performs their xG in the long term is under-performing in the short term, it may suggest that they are due a goalscoring run.
There are countless examples of this every season, most recently with Manchester United’s Rasmus Hojlund. The Dane failed to score in his first 14 Premier League games, accruing 3.1 xG. In the six games that followed, Hojlund scored seven goals from 3.4 xG.
Similarly, xG/Shot gives an idea of the quality of chances created by a team or individual. If a team has a high xG/Shot, it suggests they’re taking shots from better locations than if a team has a low xG/Shot
There are a number of ways of calculating xG, but most are taken using fairly similar criteria.
Factors taken into account include: distance from goal; whether or not the shot was taken with the stronger or weaker foot; how the shot was taken (header, volley etc.), and the action leading up to the shot (defensive error, through ball, cross etc.).
Post-shot Expected Goals, or PSxG, is an advanced measure of xG which only takes into account shots on target, and identifies how likely a shot is to result in a goal after it’s taken.
For example, a penalty typically has around 0.76 xG before it’s taken, but if the penalty is off target, the PSxG would be 0, as a penalty off target can’t result in a goal. But a penalty taken with power that goes in off the crossbar near the corner of the goal would be closer to 1.00, suggesting it’s got a better chance of going in than the average penalty.
The same applies to all shots and is a better measure of shot quality than xG, which is more of a measure of chance quality.
Expected Assisted Goals, or xAG (sometimes referred to as xA), can be better thought of as xG created. If a player makes a pass that leads to a shot, that shot’s xG will essentially be the xA of the player providing the pass.