Expected goals help more objectively assess matches. For example, a team won 2:0. They seem to be on form. However, the expected goals model shows that in terms of the threat of shots on goal, the team didn't even win by 1 goal. The rival team was more dangerous offensively, and the winner just got lucky with scoring. The forward on the other team missed clear chances on goal. In the next match, our winner once again creates a few threats, but this time they aren't so lucky: the team doesn't score and loses.
xG is part of advanced statistics. It's mainly used by football analysts and coaches. However, it's a popular model among non-professionals as well, such as fans and . xG odds can be easily found and are available for anyone to use quite easily.
bettors
Let's take a closer look at xG odds in football, including why they're necessary and how they can help choose bets.
What do we need to know about xG?
xG is an abbreviation of "expected goals". The expected goals model helps assess the risk of shots on goal and shows the probability of them leading to a goal.
The essence of the model is that all shots differ in terms of threat. For example, one shot might be made from beyond the penalty area from 40 metres away, while a second shot might be made from 11 metres. Clearly a shot from a shorter distance has a greater probability of scoring. Expected goals reflect this difference in the risk of shots on goal.
In the xG model, each shot on goal is assigned danger odds between 0 and 1, where:
- 0: no chance of a goal
- 1: 100% chance of a goal
All the odds of shots on goal are added together. The resulting number is the quantity of expected goals of the team.
xG on the example of shots on goal by Cristiano Ronaldo in 11 EPL matches in the 2021/22 season. The highest xG of two shots on goal from the penalty area is 0.84 and 0.89. This means that the probability of scoring a goal is very high. Ronaldo scored both times. However, the xG of Ronaldo's longest shot on goal from beyond the penalty area was only 0.01. Cristiano had almost no chance of scoring from this position and his shot wasn't even on goal. Source: understat.com
How is xG calculated?
Analysts assess the danger of every shot on goal. The calculation is based on a comparison of thousands of shots on goal in similar situations and the frequency of goals scored. This data is then used to form the danger odds for shots on goal.
Analysts take many factors into account. These include:
The total sum of factors analysts use to make up the model can be different. As a rule, this can lead to a small difference usually in terms of hundredths, or less frequently tenths of an xG rate.
Based on expected goals, analysts create alternative championship tables. They show where each team could be in the table if their real results coincided with what was expected.
To the left is the La Liga championship table for the 2020/21 season. To the right is the same championship table based on the expected goals calculation. Atletico has 20 points more that it would have based on the quantity of shots on goal. The club became champions, but based on expected goals it was only in 5th place. Source: understat.com
What does the xG statistic give us?
Expected goals help us reach conclusions about the teams' current performance and thus make long-term forecasts for the championship.
We can see that a club is in first place in the championship table, but in terms of xG should be in 10th. We can thus conclude that the club got so high due to accidental factors or extremely successful forward play. However, we can see that in the long-terms the results will even out, the club won't be so lucky and it will fall down the championship table.
xG models are used by sports experts, but bettors can also benefit from taking expected goals into account. The use of this factor will increase the quantity of winning bets.
For example, in the EPL 2020/21 season, Brighton and Liverpool met in the 10th round. Liverpool was in 1st place in the championship table, and Brighton in 16th. According to the championship table, Brighton was the clear outsider.
But xG showed that Brighton wasn't that bad. The team was in the top 9 in terms of expected goals. Brighton had a good defence, and its rivals had a hard time creating goal situations. In the championship table based on expected goals, Brighton was in 7th place. Individual mistakes and bad luck influenced the team's position.
Taking expected goals into account, it would be a risk to bet on a clear win by Liverpool. The final score of 1:1 confirmed this.
Taking expected goals into account, it would be a risk to bet on a clear win by Liverpool. The final score of 1:1 confirmed this.
How can xG statistics be used when betting on football?
Study the team's xG in the long term
This will help form a general impression of the team. If the team consistently scores less than shown by xG odds, then it probably has problems finalising attacks. It's risky to bet on a high individual total.
Consider expected goals when you're placing long-term bets, for example, the final position of the team in the championship table. Be careful when considering betting on teams with a serious discrepancy between expected and real goals. Sometimes this discrepancy can be explained by the team's playing style or the general performance of players, and then there's nothing to worrying about. However, in certain cases, the reason can lie in temporary factors such as raised spirits after the replacement of the team coach. Then the team's results can change rapidly.
Analyse the team's xG in matches with clubs that are similar to their closest rival
Look at how the team plays against clubs with a similar style to their closest rival. You may find a logical connection you can use in future bets.
For example, there's an upcoming match between a clear favourite and an outsider. Statistics show that when playing against outsiders, the favourite consistently scores less than they should based on expected goals.
It's likely that the favourite's forwards don't play well under pressure. Outsiders usually focus on defence and keep more players in their half of the pitch. The rival's forwards are under almost constant pressure from the defending team's players. Not every attacker is capable of taking a shot on goal in a situation like this. It's risky to place a bet on the high individual total of the favourite.
Consider the xG of the most recent matches
Sometimes a team can suddenly lose or gain form. Over the long term of the championship, its results may be stable, but may also be plagued by sudden and serious failures.
For example, over the long term of the entire championship, the team may lead in terms of goals scored and xG indicators. They seem to be on form.
But if you look at a shorter period of the last three games, the team's expected goal indicator might have fallen seriously. In fact, its indicator for this period may be no more than average. The problem is that the team's main play maker was injured. When they're not playing, it's risky betting a high team total.What are the disadvantages of xG?
xG doesn't take into account the level of the player taking the kick
The model doesn't care who's kicking the ball. It could be Cristiano Ronaldo or Anton Zabolotnii. It shows the danger of the situation based on how an average player would take the shot. If the ball is played by a player with above average technique, then they have a greater chance of scoring even in a bad situation.
Metrics don't consider attacks that don't end with a shot on goal
If a team started a dangerous attack, but is unable to complete it with a shot on goal, this attack isn't considered in the statistics.
xG can't be used independently from other statistics
The expected goals model can't replace other statistics. Using it independently of other data may lead to incorrect conclusions.
In addition to expected goals, you need to study the team's playing style and players' individual skills. You need to look at matches and assess the level of interaction between players. Look to see what sort of rivals the team plays better against. Are they attacking or defensive teams? xG is one of many evaluation tools.
In the 2019/20 season, Liverpool scored 99 points and became champion. Throughout the entire season, the Merseysiders showed significant divergence with the xG model. Taking the shot on goal danger into account, Liverpool should only have scored 74 points, which would have put them in second place.
If in the middle of the season we had only assessed xG, then we might have come to the false conclusion that Liverpool could still level out their results by the end of the championship and fall to second place.
A deeper analysis would show the reason why Liverpool is scoring more points than it should according to xG. One of the factors of Liverpool's domination was its phenomenal playing skills. Liverpool scored even from shots on goal with low xG odds, which made all the difference. Such a level of goal-scoring suggests the careful preparation of free kicks and corners more than mere luck.
Where can I find xG metrics?
All available information about expected goals is in English. There are pay-to-access sources and free sources. The most well-known sources for amateurs are understat.com, Between The Post, Total Football Analysis, and also the Caley Graphics twitter account.
Understat.com
This site keeps statistics about six championships: EPL, La Liga, Bundesliga, Serie A, Ligue 1 and the RPL. It contains championship tables with xG odds and real team indicators: number of goals scored and missed, and number of points accrued. The teams can be filtered in terms of real and expected indicators. For each match, there's a map describing shots on goal and their xG odds.
No registration is needed to use understat.com.
This is how RPL 2020/21 season looks on the site. Source: understat.com
Between the Post
Between the Post contains statistics and text analyses of club and national team matches. In addition to xG, the site contains infographics showing passes, player positions and shots on goal. Experts analyse the matches and offer detailed explanations about how the teams are playing. They also describe their strong and weak sides.
The site has two user levels. A number of match analysis articles and related statistics are available without registration or subscription. To gain full access to statistics and all the articles, you need to subscribe to the site.
Example of an infographic with xG of the national teams of Wales and Belgium for the 2022 World Cup. Source: betweentheposts.net
Total Football Analysis
There's no separate statistics section on this site. You have to look for xG in the analytical and statistical posts for each specific match, as well as articles with textual analysis.
Total Football Analysis offers partially free-of-charge content. It sometimes publishes free articles for the public. Unregistered users also receive free access to one article every month.
Registered users receive free access to three articles every month and three post-match analyses. Fully paid subscribers receive unlimited access to content.
Example of an infographic with statistics from the match between West Ham and Liverpool. Source: totalfootballanalysis.com
Caley Graphics
Caley Graphics is the account of Michael Caley, a football analyst. In his post-match feed, he publishes shot-on-goal maps with the total value of expected goals. Calley covers the most popular championships, such s the EPL, Bundesliga, League of Champions, and UEFA. Access to Caley's feed is free-of-charge.
Caley uses statistics from Opta, a British analytical company. Source: twitter.com/Caley_graphics
What abbreviations are used in xG models?
- xGA: expected goals allowed. This shows how many goals a team is expected to concede based on the threat of their rival's shots-on-goal. This allows for the team's defence to be evaluated.
- xPTS: expected points. The odds take into account the threat of a team's shots on their rival's goal and shots allowed on their own goal. Then it calculates how many points the team could have gained on the basis of the quality of the positions. This gives an idea of the team's general form. It also allows for the playing level of football clubs to be compared.
There are also more in-depth metrics. Here are some of them:
- NPG (non-penalty goals): expected goals not taking penalties into account. This helps a team's attack to be more precisely evaluated. Penalties are usually given high odds, 0.7+. For this reason, shots from 11 metres strongly impact the total xG. If a team regularly scores from penalties, this might give the false impression that it frequently creates goal opportunities.
- PPDA: team pressure. This shows how many passes a rival has made before the team tries to take the ball away from them. High odds show that the team is under less pressure and less often plays defensively.
- xA is expected assists. This is an individual indicator. It shows the importance of key passes and how many goals could have been scored after a player's goal pass. The odds reflect the player's creativity.
- xGBuildup: the significance of the build up and attack development. This shows the commitment of the player in threatening attacks that don't end in a shot on goal. The odds help distinguish players who made a significant contribution to the team's attack.
- xGOT: expected goals on target. The difference between xG and xGOT is that the latter shows the danger of the situation after the shot has been taken, not before. This indicator considers the player taking the shot. For example Harry Kane's xG is 0.03. However, Kane is a top finisher and he's capable of scoring even in a very bad situation. Therefore, the xGOT of his shot is 0.54.