What Are Expected Goals (xG) in Football?

Data and statistics have become an inherent part of football, with analysts decomposing every aspect of the game into the prime factors. Managers, pundits, sports bettors, even fans – all look and analyse stats. And while their reasons differ, the primary goal remains the same for everyone; they all want to understand better how the beautiful game works. 

However, there’s one statistic that has gained a lot of popularity in the past few years – the famous expected goals (xG for short). First used by the betting community, the metric has quickly been adopted by mainstream broadcasters, pundits, and fans. This shouldn’t surprise as expected goals are indeed one of the most advanced and eye-opening metrics in modern football. 

The only problem with xG is that not many understand what it means. If that’s the case with you, don’t worry. We’re here to help. Below, you’ll find a comprehensive guide to what expected goals are, how the stat is calculated, and how you can take advantage of it.

Understanding Expected Goals

Simply put, the expected goals metric is used to calculate how many goals a player or a team should’ve scored in a game. xG takes into account several variables from before the shot was taken, combining them with historical data to measure the quality of a chance and its likelihood of conversion into a goal.

In other words, xG can be understood as an average probability of an opportunity being scored on a scale from zero to one. The higher the value, the better the chance. Okay, but how to read this metric? Suppose a certain opportunity was given 0.4xG. This means that from ten practically identical shots, an average player should score four goals (40% conversion rate).

Of course, a player or a team can outperform or underperform this metric. After all, football is a game that often defies the laws of logic. That being said, you can expect a player to convert a 0.1xG chance and miss a 0.9xG shot. That’s just the beauty of football, though. At least as long as it’s not your favourite team missing the chance, of course.

Origins of the xG Model

The idea of expected goals emerged in 2012, where Sam Green from the leading sports statistic company Opta, inspired by similar models used in American sports, first introduced his approach of measuring the performance of English Premier League goalscorers. 

In 2017, the term ‘expected goals’ entered the mainstream media, debuting in the BBC’s Match of the Day. And nothing has been the same since.

To create their model, Opta, led by Sam Green, analysed approximately 300,000 shots from all of the top-tier football leagues, dividing them according to numerous variables, such as the angle of the shot, distance to the goal, assist type, etc. This approach allowed them to assign an xG value to every attempt, determining how good each opportunity is. What’s more, Opta continues to collect data, ensuring that their model is continuously refined. 

Also worth mentioning is that there isn’t just one expected goals model used at the moment. Numerous football analysts and statistic enthusiasts have developed their own variations, adding or excluding variables. The principal idea, however, remains the same, with each model calculating the likelihood of a shot being scored.

How Is the xG Value Calculated?

All football fans can easily distinguish between a great, average, or impossible chance to score a goal based on such factors as proximity to the goal or angle of the shot. Converting those observations into numbers, though, is a different story. This is where expected goals come into play.

By combining hundreds of thousands of shots from their historical data and combining them with several variables, Opta created the model that calculates the likelihood of a scoring opportunity being converted. Among the essential factors Opta and other sports statistics companies use to determine xG, we can include:

• Distance to the goal: the closer a player is when taking their shot, the higher the xG.

• Angle of the shot: more acute angles are harder to score from, lowering the xG value.

• Body part: xG can vary depending on the shooting body part; for instance, headers are more challenging to score than shots taken with a preferred foot.

• Assist type: how the chance was created also affects xG (a through ball, cross, etc.).
• Passage of play: xG varies depending on whether a shot comes from open play, counter-attack, corner kick, direct free kick, etc.

As briefly mentioned, there are many different models used to calculate xG. Some are more complex than others, considering more advanced variables, like the scoring quality of the shooter or the opponent’s defensive play.

Limitations of the Expected Goals Metric

Although fairly accurate, you should take expected goals values with a grain of salt. That’s because the xG model is only as good as the variables it takes into calculations. Unfortunately, even the more advanced models don’t consider such factors as a shot power, whether a goalkeeper is off balance, or weather conditions. 

Who shoots and defends also makes a massive difference. For instance, suppose Robert Lewandowski and Harry Maguire taking an identical shot. It’s obvious that it’s Lewandowski who’s more likely to score. Well, according to xG models, they have the same likelihood of converting that chance into a goal. 

That’s because xG models deal in averages. Those, given the randomness of football matches, aren’t always correct. With this in mind, you should treat expected goals as supportive information rather than a certain indication of how many goals a player or a team will score. 

Misconceptions Regarding Expected Goals

The problem with expected goals stats is that many people in the football community often misuse this metric, applying it to wrong scenarios. The most common misconception about xG is that a team with a higher xG should’ve won a game. That’s simply not true. Expected goals are used to measure the quality of a chance and not predict the game’s outcome.

For instance, imagine a situation where one team takes an early lead. In this situation, they don’t really need to push and create chances. They can just sit back and wait for the opportunity to kill the game. On the other hand, the other team has to attack and generate more chances, often translating into a higher xG value.

Another misconception comes from the already mentioned limitation of xG models only dealing in averages. As football fans know, each game is a separate story, and there are tons of random events that can disrupt the final outcome. As discussed, xG models don’t take those into account, which many people don’t understand, blindly believing that a 0.5xG guarantees a 50% chance of a player scoring. As you already know, this doesn’t work like that.

Using xG Data for Betting

Although they come with several limitations, expected goals can prove advantageous when it comes to sports betting, providing bettors with information a final score may not reflect. That’s because football, as a low-scoring sport, often generates outcomes that can be misleading. 

For example, you’ve probably seen many games where the dominant team lost despite creating more chances or having higher ball possession. In that case, a final score doesn’t really tell the full story. The number of shots is also not really representative, as every opportunity is different.

Instead, you should look at the xG data. Using it, you can get a better idea of how a team is actually performing in terms of creating and converting chances. You can use this information to your advantage regarding the upcoming features. For example, a team that has been underperforming recently is likely to return to their average xG value and vice versa.

Key Takeaways

‘Expected Goals’ is a concept introduced by Opta analyst Sam Green in 2012. It refers to the likelihood of scoring a goal from a specific chance, based on the historical data and on-pitch variables, such as proximity to the goal or type of assist. 

And while xG models aren’t perfect, they provide a great insight into how a player or team performs in terms of creating and converting their chances. This can come in handy when analysing a team’s or player’s performance, proving to be an excellent aid for managers, pundits, and bettors alike. 

Hopefully, after reading this guide you get a better understanding of how to read xG and how to use it to your advantage. Be it sports betting, or discussion with your fellow football fans. 

FAQ

Where can I find xG data?

Since calculating xG yourself is impossible (well, maybe not impossible, but we’d love to see how you pull it off), you need to do some research to find reliable data. Fortunately, there are many websites you can look at to get the information you need. Just type in the xG data you want to analyse in your browser.

What are the other similar metrics in football?

Football analysts are an inquisitive bunch. Nowadays, you can find a stat for practically anything involving the beautiful game. However, when it comes to metrics similar to xG, you might come across such stats as expected assists (xA), non-penalty expected goals (npxG), expected goals against (xGa), expected goals for (xGf), or expected points (xPts).

What does an xG value tell me about a team or player?

In general, a team or player with a higher xG value should score more goals. And while it’s not always the case, players and teams can underperform and overperform, disrupting their average xG.