If you’re the kind of punter that simply looks at league tables, current form, and team news to formulate bets then hey, there’s absolutely nothing wrong with that and you can enjoy long-term profit or big acca wins using that information.
But like the best detectives, there is a wider range of stats and data we can use to supplement the ‘basics’ of betting, and actually can give us a much better understanding of a team or a league.
We have all watched football games where we’ve come away thinking ‘how the hell did Team A win that?’. They may have been soundly beaten in all aspects of the game, barely touched the ball, and yielded countless chances, and yet – football being football – they’ve managed to come away with a sneaky 1-0 victory.
And there is always the flipside: Team B, who have played some excellent football but go home empty-handed.
You might just brush that off as football just being ‘a funny old game’, but what happens if Team A are being ‘lucky’ on a consistent basis? What if Team B continue to perform well without getting the results they deserve?
Eventually, the tables will turn, and Team A will get their comeuppance while Team B will start putting good results together.
And this, essentially, is what Expected Goals is about; that weird stat thing they use on Match of the Day and in the papers.
What is Expected Goals?
Expected Goals (xG as it is sometimes displayed) is a descriptive measure of outlining exactly what happened in a football match.
The layman may only be interested in results and scorelines, but as we know that’s not always the best way to determine who had the better of a particular 90 minutes.
Expected Goals does that more effectively, simply by showing numerically who played better in the department that is truly underrated when analysing football matches: the quality of chances created.
How is Expected Goals Calculated?
Every shot at goal and chance created is given a numerical value based on the likelihood of said chance ending up with the ball in the back of the net.
This is dictated by historical data; a pitch will be broken down into quadrants, and data from hundreds if not thousands of matches is used to calculate the probability of scoring with a shot from each location.
But xG is smarter than that, in that other factors – quality of pass received, the number of defenders in close proximity, positioning of the goalkeeper – are also introduced to give an extremely accurate measure of how likely the ball is to be put in the old onion bag; hence the name ‘Expected Goals’.
Below is a very crude image that explains the concept of xG visually:
A shot from the area marked X on the pitch has a high chance of ending up in the net; depending on factors such as defensive positioning, quality of pass/cross etc already outlined.
A long range shot from Y has a lower rate of success, while you can count on one hand the number of times per decade a shot from Z somehow finds its way past the goalkeeper.
All of the X, Y and Z’s for a particular game are accumulated, and so you might see a final Expected Goals result displayed something like this:
Arsenal 1.73-1.13 Liverpool
How Can We Use Expected Goals in Our Betting?
Here’s the fun part. By tracking the Expected Goals data, we are able to get a better picture of exactly how a certain team is performing.
Teams that are consistently ‘lucky’, i.e. their results often outperform their xG data, will presumably somewhere down the line suffer a downturn in form.
Conversely, ‘unlucky’ teams will, you would think, enjoy an improvement in results if they maintain their excellent xG data. After all, this is a more accurate reflection of a team’s ongoing abilities.
When it comes to placing your bets, there are two ways in which we can leverage Expected Goals numbers to our advantage:
Sometimes, it is worthwhile breaking down a team’s form in blocks of, say, four games.
This can often be a better reflection than the league table: maybe players are out injured/suspended (or returning from a spell on the sidelines), perhaps a failing manager has been replaced, maybe they have lost players to competitions like the African Cup of Nations.
Collecting game-by-game xG data gives us a view of how efficiently a team is attacking, how well (or otherwise) they are defending, and how confident they are when heading into battle.
Remember, football teams take a lot of positives from good performances; even if they go down to unlucky defeats.
And you can even check xG figures for home and away matches, which will give you a greater understanding of how teams approach games in front of their own supporters and on their travels.
It really cannot be stressed enough that xG data has to be used in your weekly bets, alongside team news and other variables of course.
As many of you will be aware, a number of outright betting markets don’t close until well into the season.
From league winners to top four finishes to relegation, monitoring longer term xG offers an insight into how well a team is likely to do in the next few months.
Remember, runs of good or bad fortune will even themselves out, and ‘better’ teams, i.e. those who enjoy overwhelming xG supremacy over their opponents on a consistent basis, can be backed to do well and vice versa.
The xP Model
Once we have our Expected Goals data, we can manipulate it any way we see fit. We like to use our own exclusive xP model – xP stands for Expected Points, which we can calculate from our xG data.
We say ‘exclusive’, there are plenty of other folk doing something similar, but ours uses an xP per game approach. It’s a personal thing really and you can create your own in a way that suits you.
Our model is simple: where the xG count is greater than 0.25, the team ‘winning’ takes three points and the loser gets zilch. If the xG differential is 0.25 or lower, then we class that as a draw. Then, we dish out the points as normal and produce a hypothetical league table based on the data.
Here’s how the Premier League table looked early in November 2018:
On the left is the overall XPPG (Expected Points Per Game) table based upon individual G collections for each game.
In the centre are the XPPG splits for home and away matches.
And on the right is a form guide based on individual xG scores for each side’s last four fixtures. The numbers are the xG count, and the colours denote the result (green = win, yellow = draw, red = loss).
To reiterate, this is simply our interpretation of data that is freely available online; you can use the figures as you please.
Our system, we like to think, ticks both the short and the long-term betting view. The form guide on the right indicates how a team is doing right now, while the tables to the left give a wider scale look at what’s occurring.
Where Can You Get Expected Goals From?
If you wanted to collect your own Expected Goals data, it would take an eternity: you would have to watch hours and hours of games each week, and manually record the data for each shot taken. Let’s face it, very few of us are able to justify that.
Happily, there are lots of big brands and smaller data firms that collate xG numbers, and some of them even publish these, for free, on the world wide web.
If you regularly use your mobile phone then we heartily recommend you download the Infogol app. They have game-by-game xG data for the Premier League and other top divisions across Europe, and publish league tables based on actual points and adapted Expected Points standings.
If you operate more on a laptop, then point your browser towards the Understat site. This publishes xG scores from across Europe, and has a fantastic player-based xG and xA (Expected Assists) table that will help those who bet on the player prop markets and/or if you have a fantasy team.
Finally, Expected Goals data for the lower leagues in England is harder to come by, but you can check out Ben Mayhew’s excellent Experimental 3-6-1, where you can find weekly xG data for English League One and Two, as well as other cracking bits of analysis and visualisation.