One of the most important aspects of betting is using data to your advantage. Football can be a tough game to call and you have to be prepared for results not to go your way once in a while, but using data to inform your bets should make you money long term.
The depth of the data that we have at our fingertips these days is substantial, to say the least, and one of the latest variables to come into play since around 2018 onwards is that of expected goals. This is basically a prediction of the number of goals that each team will score in each match based on previous performances and other factors.
The stat is a really useful one as it not only allows you to see the number of goals that each team are expected to score, but it also allows you to get creative with that info, taking these goals and then applying them to other betting markets to create better betting options.
What is Expected Goals?
First thing to note is that Expected Goals is often referred to as “(xG)”, so if we do the same, you know what we mean.
The Expected Goals is stat that takes a huge range of data over the course of the season to work out the number of goals that each team are likely to score or concede in each match. They look at player performances from each team and from which position or scenario a goal has occured, then based on a huge amount of data state the number of goals they should have scored, but not necessarily what they have scored.
There are actually a number of variations of this stat and each takes the data and applies it a little differently than the other.
- Expected Goals (xG) – This is the original statistic and it works out the projected number of goals that each team will score in the game based on the quality of their shots on or off target.
- Expected Goals per 90 (xG/90) – This is the statistic for expected goals scored for each individual player over 90 minutes.
- Non-penalty Expected Goals (npxG) – Includes expected goals, but removes penalties from the overall statistic.
- Expected Goals for (xGf) – The number of goals that the individual team is expected to score.
- Expected Goals against (xGa) – The number of goals the team is expected to concede.
- Expected Goals assisted (xA) – This stat refers to the number of assists each player is expected to make in the match.
- Expected points (xPts) – This is the number of points that each team should statistically win over the season. It will also create an expected finishing position for that season based on the results.
How is Expected Goals Calculated?
You may have noticed that we’ve mentioned “data” a lot already, so we wanted to break down how exactly this data is formulated and then amalgamated to make Expected Goals.
The Expected Goals number is based from 0 to 1. The higher the number, the higher the likelihood that the player or the team will score a goal. For example, if based on the data the player is likely to score 50% of the time, then this will be represented as 0.50.
After a game, the total number of shots from that match are brought together to show the Expected Goals that should have been scored. This is based on an algorithm that takes dozens of factors into consideration, but more on that later.
For example, if Man United were playing Chelsea and had 10 shots during the game it might look something like this, with the algorithm showing the probability of each shot becoming a goal:
|Shot 1||15% (0.15)|
|Shot 2||30% (0.30)|
|Shot 3||2% (0.02)|
|Shot 4||45% (0.45)|
|Shot 5||73% (0.73)|
|Shot 6||14% (0.14)|
|Shot 7||5% (0.05)|
|Shot 8||81% (0.81)|
|Shot 9||22% (0.22)|
|Shot 10||41% (0.41)|
All you need to do then is add up the Expected Goal values for each shot and you get the overall Expected Goals for that match. In this case it would have been: 3.28 Goals.
The data is actually created by Opta, who are a British sports analytics company and have more statistics and bits of data than you could ever imagine. They analyse all of the higher league English matches and even some of the lower leagues a well. They are the ‘Stato’ of the sporting world.
The data that they have used is been based on over 300,000 shots. They then analysed each of these shots and determined what needs to be done for that goal to be scored and what events have led up to that goal in order to make it happen. This might include things like the distance from the goal, angle of the shot, type of goal, the number of passes prior to the goal and so on. The one thing that is not taken into account is player ability.
To explain this a little better, let’s say that a goal has been scored that started with a pass from the keeper out wide to the right winger. The right winger beats one defender, whips the ball into the box and a player scores from 6 yards out with his head. Let’s assume that the exact same scenario happens again, but this time the player fails to score. It would mean that 1 in 2 chances of this type have been successful, so the Expected Goal value would be 50% (0.50).
That means that if the exact same scenario happened again, the player would be expected to score 50% of the time, with a xG of 0.50.
Interpreting the Data
You’ll have noticed that the data that Expected Goals represents is fractional, yet there is no such thing as a fraction of a goal. This means that you need to interpret the data in a way that will gain you leverage when betting.
As football is only ever played between two teams in a single match, you should compare the data for one team with the opposition. So, you might have an Expected Goals value of 2.50 and 1.80 for the two teams respectively. Obviously, they can’t score 2.50 and 1.80 goals, but you can see that one is higher than the other. This shows that you statistically want to be taking the team that expected to score 2.50 to win the game, as they are forecast to score more than their opposition.
Applying Expected Goals to Betting
Expected Goals is a very useful tool that can be put to good use as part of a betting strategy. We want to emphasise ‘as part’ because it’s not perfect and you really need to be looking at more data than just this to form your bets. However, here’s why we think it’s a strong place to start.
We really like that Expected Goals gives a better outlook on how teams are performing than just relying on form alone. Sometimes football can be a cruel game, and you can create a boat load of chances only for the opposition’s keeper to have a blinder and them to steal a goal.
Expected Goals allows you to see when someone is performing better than the results, and this means that you can anticipate a turn in form. The league position is one area that we like to look at.
Here is a game from Brazil between Gremio and Chapecoense. As you can see, Gremio are actually languishing somewhat at 13t in the league, but based on Expected Goals they should be in 7th with a bit more luck. They are still strong favourites against Chapecoense, but if they were playing a more dangerous team, then you could argue a case for them being a better price than they are based on league position and past results alone.
You could even back this up with data from the Expected team goals, showing that they are currently averaging just 1.08 and they are expected at 1.49 given the chances they are creating.
Once the season really kicks in, you can start breaking down the Expected Goals data into both home form and away form. This is vitally important for football betting as you often see sides set up very differently at home than they would be away from home, so the overall numbers can be a little off because of that.
As more games are played, you can start to look at other markets that might offer some value. We’ve spoken a number of times in other articles about how trying to predict the correct score for the match can really open doors when betting. You don’t necessarily have to bet on the correct score, but building an idea of what it could be opens up so many other markets that you can then access to find potentially better value bets.
For example, if we stick with the Gremio game above, we can see that the expected goals for the game (total) is 2.75 goals overall. We also see that the game is expected to be quite tight, even though Gremio has an advantage.
Gremio priced at 1.30 to win the game seems very short considering all of this.
Both teams are averaging over 2.50 goals per game as their Expected, even though Gremio is below this as an “actual” figure. The over 2.5 goals bet at 2.10 seems to stack up based on the expected number of goals that will be scored and offers much better value than a Gremio win.
This is just one example of how you might use the data to choose a bet, but it demonstrates the importance of using Expected Goals to open up other markets. You can also look at individual players for markets such as 1st or anytime goalscorers. Check out their actual and expected goals, and then compare them overall to see if you can find any value.