Expected Goals Explained - The Ultimate Guide To xG (2018 Update)
In this article we explain all you need to know about Expected Goals (xG) including exactly what it is and the different variations, how to calculate it, how we can apply it to sports betting and more.
Expected Goals (xG) Explained
Advanced Metrics (the term used in relation to the analysis of sports to measure in-game productivity and efficiency) are already utilised within many sports around the world – most notably Baseball, Basketball and American Football. Now, they are making their way into mainstream football, in the form of Expected Goals.
What Is Expected Goals?
‘Expected Goals’ (symbolised as xG) is a measure, usually expressed as a number between 0 and 1, on whether a given shot will result in a goal. It takes into account a range of factors and historical data and allows us to identify how many goals a player or team should have scored based on the quality of chances they had during a game.
For a visual explanation of expected goals you can watch this brilliant video by the awesome guys at TifoFootball.
Expected Goals Glossary
There are many variations of Expected Goals, these include:
- Expected Goals (xG) – The volume of goals that either a player or team will be expected to score based on the factors that a model takes into consideration.
- Non-Penalty Expected Goals (npxG) – The total expected goals minus any expected goals from penalty attempts.
- Expected Goals For (xGf) – The amount of goals a team is expected to have scored based on the expected goals data.
- Expected Goals Against (xGa) – The number of goals a team should have conceded based on the expected goals data.
- Expected Goals Assisted (xA) – The total number of assists a player should have produced based on expected goals taken directly from their passes.
- Expected Points (xPts) – The number of points a team is expected to have won in correlation with the expected goals data.
Expected Goals In More Detail
An xG of 1 is the highest value a single shot can be, which implies that a player has a 100% chance of scoring. The higher the value of the xG, the more likely the player is to convert the opportunity.
The use of npxG is particularly useful as it provides a more accurate analysis. Since penalties have an xG of 0.76, they can significantly distort both a player’s and team’s expected goals. Since the penalties may not have been earned or deserved, they can provide an inaccurate look to the data. There is a good article here that goes into more detail on Expected Penalty Goals.
Prior to expected goals, statistics such as ‘Total Shots’ and specifically ‘Shots On Target’ were used when analysing a match and similarly to a final score line, they can be deceptive when considering that a shot with an xG of 0.13 classed as the same as a shot that has an xG of 0.83.
For example, let’s imagine that Team A took a total of 17 shots during a game while Team B only took 8. From these stats, we would be under the impression that Team A deserved to win. However, if we looked at the expected goals data, we would see that Team A had an xG of 1.34 from the game while Team B had an xG of 2.18.
One popular criticism of the data is that current models do not take into consideration the talent levels of:
- The player shooting.
- The player in goal.
This is something that will obviously influence the xG value when it is implemented and is therefore something to keep in mind.
How Is Expected Goals Calculated
There are a range of different models used to measure expected goals, ranging from the simple to the complex. For example, Opta, the world’s leading supplier of sports data, analysed over 300,000 shots to help create their model.
Below is a video of Opta’s Duncan Alexander discussing the xG metric and how it can help us better understand team and player performance.
The variables that are taken into account when determining the likelihood of a goal include:
- Distance from goal – Generally, the closer you are the higher the xG.
- Angle of the shot – Generally, the more acute the angle, the lower the xG.
- Shooting part – Was it with the strong foot, weak foot or a header?
- Passage of play – Was it from open-play or from a set piece?
- Chance creation – Did the opportunity come from a cross, a through ball etc?
- The shot – Was it from a rebound, did it come after beating an opponent etc?
While these provide a good standard for expected goals analysis, some of the more complex models also take into consideration factors such as the defensive play of opponents. Defending is just as big of a factor in a game as attacking is, so by taking it into account, the data is likely to be more reliable.
How To Apply Expected Goals To Sports Betting
Expected goals data is advantageous to sports bettors as it provides information that a final score may not always reflect.
Football is generally a low scoring sport and as such, goals come in a small commodity, meaning the final score of a game can be misleading.
For example, you may see a team dominate a game in possession, territory and chances created, yet somehow still manage to lose. The basic goal data (final score) will therefore be unrepresentative of the game and thus can’t be used to form an opinion on future fixtures.
Nevertheless, what can be used for future purposes is the xG goals data that has come from the game. Using this data we can remove any perils about the likelihood of finishing at both ends of the pitch and get a more reliable interpretation of a team’s overall quality.
With regards to upcoming matches, expected goals data can help us identify value. If a team has been over-performing or under-performing their xG metric, they are likely to soon return to their average.
For example, let’s imagine Team A has picked up only 1 point from their last three games despite comfortably beating all three of their opponents on the expected goals data. Due to this poor run, Team A are priced at greater odds to win their next fixture than what the data suggests. This would represent value.
While xG data can and should only be used as a guideline, if it supports your research and you believe a price is deemed value, then it is likely to be a good bet.
There is always money to be made in ante-post markets and by using a system rather than going on gut instinct, you are more likely to be successful.
While you can use the expected goals data to predict upcoming matches, it can also be used for forecasts, such as table standings and golden boot standings.
By using xG goals data both for and against from previous campaigns, we can create an alternative league table, which provides an informative display of how the season went and can help us predict future performance.
However, when using this data for future predictions, it is important to remember that these statistics do not take into consideration factors such as transfers, injuries, form and new managers.
In essence, expected goals is a way of assigning a ‘quality’ value to every goal scoring opportunity, based on the information available.
There has been a serious amount of growth in the modelling of xG and as time goes on, the more data that is collected, the more reliable and accurate the metric will become.
It is important to remember however, that the analysis is not always 100% representative of a situation and there will therefore always be outliers.
While the utility of expected goals has, does and will continue to divide opinion, what is for sure is that it is here to stay.