If you listen to the podcast or follow me on twitter, you will definitely have heard the term "expected goals" being thrown around. Over the past year or two, expected goals ("xG") has become one of the more popular statistical tools used to analyze a soccer match. Despite its usefulness and rise in popularity, there has still been a lot of pushback; plenty of pundits, fans, and executives hold to a more traditional understanding of the game, one that has no room for statistics and only looks at results (this rant from Craig Burley is pretty epic). A lot of this hesitation comes from several misunderstandings.
Statistics and xG aren't a substitute for watching the game. Analytics are a way to quantify what we see on the field, but the eye test is still an incredibly valuable tool. In-person scouts are still the most reliable way to judge a player. Watching a team play will most likely help a manager game plan more than looking at stats. But looking at statistics can help provide depth to what we see on the field. It can help us understand why certain results happened. It can help us further analyze issues teams are having. All of these are useful tools. But at the core, statistics are a companion to what we see on the pitch.
xG is no different. It can help clubs analyze their play or allow pundits to back up their points. It's not a substitute for the human eye.
On the flip side, statistics like xG are a helpful way to stay objective. It’s very easy to get caught up in the emotion of a game. Our memory is a tricky thing, and we tend to let things like stadium atmosphere and results shape our opinion of a team’s performance. Statistics help us see how the team actually performed, how well individual players are actually playing from an objective point of view. This isn’t to say that atmosphere or results shouldn’t have an impact on our opinion of a match, but it’s about finding a balance.
With that lengthy caveat out of the way, what exactly is xG? At the most basic level, xG is the number of goals we can expect to be scored based on where and how each shot was taken. It's pretty simple. You're more likely to score with your foot from inside the 6-yard-box than on a header from outside the 18.
For example, this tap-in from Ropapa Mensah had an xG of 0.453, meaning this type of shot goes in 45% of the time (goal happens around 0:15 in the video).
Conversely, Akinyode's rocket against Cincinnati (2:51 in the video) had an xG of 0.054 - a shot from that location has about a 5% chance of going in.
There are other factors that come into play when calculating xG: whether the shot was taken from a set piece or from open play, whether the ball was crossed or played on the ground, whether it was a penalty kick or a counter attack. All of these scenarios are factored in to determining the odds of a given shot going into the net.
This is really helpful for us to have a more detailed understanding of the game. Simply looking at shots taken doesn't give us an accurate picture of what happened on the field. Let's take Nashville SC's 2018 season as an example. Nashville didn't score enough. That was widely documented and pretty obvious from watching the team play. However, the xG shows that it was somewhat unfair to label them a bad attacking team. Nashville was pretty average from a chance creation standpoint, but only managed to score 42 regular season goals. No Eastern Conference playoff team scored less. But according to xG, Nashville should have scored 54. The issues with goal scoring was less of a systemic issue and more of an issue of poor finishing at the end of attacking moves. Don't get me wrong - there were plenty of systemic and tactical issues, but at the end of the day, players have to finish chances, and they didn't.
This also helps us analyze individuals performances. Just because a player is at the top of the scoring charts doesn't necessarily mean he or she is the best finisher. It could just mean that the player is in a system which creates a lot of relatively easy chances. Players like Harry Kane, on the other hand, have consistently over performed their xG, finishing chances that don't typically go in. For Nashville last season, Tucker Hume scored seven goals, even though his xG was only 4.74. It's not enough just to say "this player doesn't score enough" or look at how their conversion percentage (how many of their shots result in goals) - we need to look at the actual quality of chances he/she gets in order to determine how well they're finishing.
In addition to helping analyze past results, xG can be used to project how certain matches will play out by comparing the chances they are creating and the chances they allow. Obviously, predictions are at best educated guesses and anything can happen on the pitch, but it does allow us to go into a game with more context and a better idea of how the teams match up.
A common misconception is that winning the xG battle means you should win the game. Eliot McKinley does a great job explaining this here. Basically, xG gives the most probable outcome for goals scored in a game, but the number of shots taken have a major impact as well. The home team could take 15 shots from 30 yards and end up with an xG sum equal to the away team, who took 2 shots in the 6 yard box on a counter attack. In reality, the away team is more likely to score in this scenario, despite the teams being even on xG. So there is a nuance to xG that needs to be accounted for.
Unfortunately, in depth xG stats can be tough to access for USL. Websites like American Soccer Analysis do a brilliant job with their xG coverage of MLS, and Understat has comprehensive data for Europe's Top 5 leagues, but this type of detailed coverage is a luxury that doesn't yet exist in USL. As we did last year, we'll continue to provide xG stats for each Nashville SC league match.
Expected goals is a really interesting way to analyze the game we love, and can provide a more in-depth way to understand what we see on the field. The more informed we can be, the better.