What are Expected Goals (xG)?
Expected goals calculates how many goals a team should have scored based on the quality of the chances created.
By comparing actual goals to expected goals we can see whether a team has been lucky or unlucky in any given match. For example, the champions league final between Liverpool and Spurs had xG of 1.3 for Liverpool and 1.2 for Spurs. As such, a 1-1 draw would have been a fair outcome. It can be concluded that Liverpool were lucky to win so comfortably.
The process for calculating an expected goal from any given chance was done by Opta. They reviewed hundreds of thousands of historical shots to then work out the percentage chance of a shot being scored from any particular situation.
Close range vs long range shots
In simple terms, shots from close range and in front of the goal tend to have the highest percentage chance of being scored. The best example is a penalty which is on average scored 76% of the time. That means that if a team were awarded one penalty and no other shots, their xG would be 0.76.
Long shots from outside the area have a lot smaller percentage chance of conversion and usually fall in the 1% chance of a goal bucket.
Using the same example, a team with one penalty versus a team who has taken 20 long range shots should on average win over the long term i.e. xG of 0.76 versus 0.2 (roughly 1-0). This is a good example of how shots can provide a misleading impression on how a game is going.
Expected Goals (xG) to predict future matches
The main strength of Expected Goals (xG) over shots is there is a lot lower variability (small standard deviation) in the xG results. Teams are fairly consistent in the xG created and conceded while shots can vary significantly game to game. Due to being more consistent, xG are a more useful tool in predicting future results.
Predicting match results
This site uses Expected Goals and multiple other factors (142 in total!) to predict the results of future matches.
The predictions are then converted to a percentage chance of winning. These percentages are then converted into odds to see if there is value in the available match odds.
Lower standard deviation
History does not always predict the future
As with everything, Expected Goals comes with a number of limitations.
Expected versus actual
While Expected Goals are a good predictor of a result, actual goals do not always align to the expected goals. The Champions League final being a prime example. However, over the long run, the number of goals a team scores tends to revert to the Expected Goals. If a team is scoring a lot more than the Expected Goals, there is a good chance they are overachieving. Therefore they could be a good option to lay in future games, especially if they come up against the opposite situation – a team who is creating lots of good chances but not putting them away. For example, laying Man U towards the end of their good run of form (after OGS was appointed).
Player skill levels
The other main limitation is Expected Goals does not take account of the player taking a shot. As any football fan will know, all players are not created equal. For example, Vincent Kompany’s goal against Leicester which turned out to be crucial in the 2019 EPL title race only had a 1% chance of going in. As this % chance is an average of all types of players, the true odds were probably a lot lower. If Kevin De Bruyne was taking the shot it would still be classed as 1% but the true chance would probably be closer to 5%. Again, over the long run Expected Goals should revert to actual goals but there can be short term fluctuations.
Watch the match
Trying to predict the future purely based off history is not going to provide perfect results every time. While xG provides a good indication of which team should be on top, I advise people to watch the match before putting on a trade. This provides crucial insights into whether the team we expect to win are actually playing like their form suggests.
The below sections provides some further analysis and answers a number of common questions.