It is very much the zeitgeist term in modern football analysis, and not everyone has got to grips with Expected Goals, or xG as it is commonly abbreviated to. Indeed, Jeff Stelling’s less than rapt opinion of the data model went viral in the past couple of weeks.
The reaction of known progressive thinkers Paul Merson, Phil Thompson and Charlie Nicholas speaks volumes!
Everyone will have their own opinion on xG, but it is clear that an obvious lack of education about the model is hampering its use in the media and also its wider understanding as a tool that adds context to the beautiful game without offering up concrete explanation.
What is Expected Goals?
For the uninitiated, the Expected Goals model attributes a decimal value to each chance created in a football match. So if a player is presented with a tap-in in the centre of the goal then that will naturally be given a high xG factor. Conversely, an overhead kick from the corner flag will not.
At the end of 90 minutes, these pieces of data are added up to create an overall Expected Goals picture, so if a game ends Manchester City 3.10-0.35 Brighton we get a decent understanding of how the match panned out.
The bit that Stelling and other naysayers struggle with is why does this data matter? If the example above ended 3-0 to Man City in terms of the actual result, then the xG data is wholly irrelevant. Even if Brighton won 1-0, the Expected Goals count is still of little value.
That is true of course: the only currency that matters in football is sticking the ball in the back of the net. But they key to Expected Goals is providing context – it unravels the story within the story that every 90 minutes of action, from the most exquisite example of the beautiful game to the most tawdry bore-fest, serves up.
So let’s take that thinking and run with….collecting xG data across the whole season. What would we expect to find? Well, we might assume that the team that finishes top of the Premier League has the best Expected Goals ratio, and the team that finishes bottom would have the worst.
If only it was that simple….
The beauty of xG is that it identifies what we might term ‘outliers’, or teams that are wildly under or overachieving according to the data. There are two exotic examples of that in action from the 2017/18 campaign already.
The Dyche Effect
Later in this article we will publish an Expected Goals table covering the Premier League this term, but for now we can tell you that Burnley – sitting pretty in fourth in ‘real life’ – occupy a rather more modest eleventh according to our findings.
Why is that? Simply, their opponents are routinely creating more and better chances than the Clarets on a game-by-game basis.
But how can that be so? Sean Dyche’s men are challenging for a Champions League place based on their actual results. But again, and as we must reiterate, context is king: the season is 17 games old, and up to press Burnley have been routinely outplayed more often than not.
The doom-mongers of Expected Goals will use the Clarets as the working example of why the model is hokum, but just remember this: as smart punters, we use xG as a guide for our long-term thinking. Weak Expected Goals data does not indicate that Burnley are going to lose their next game, but it does offer a suggestion that their current fourth place will be tough to sustain if they continue to create fewer ‘big’ chances than their opponents.
Eagles Set to Fly?
When we minimise the data to cover the last seven matches, we see that Crystal Palace sit comfortably in sixth in our own modified Premier League table.
That indicates a vast improvement from the Eagles; backed by the all-important context of Roy Hodgson’s appointment, the return from injury of Wilfried Zaha and the excellent form of Ruben Loftus-Cheek, as well as an agreeable fixture list.
Nevertheless, Palace remain in the relegation zone of the actual table, even though a points haul of ten points from their last six outings is a ratio that would comfortably see them stay up.
Using the xG model as our guide, we can conclude that Hodgson’s men should easily avoid the drop this term – a fact generally overlooked by the bookmakers, with Stan James offering odds of 6/4 on Palace to stay up. Come May, they could be made to look very foolish and our bank balance could be inflated as a result.
Using Expected Goals (xG) in Betting
Football betting is fascinating, because this is a low-scoring sport where individual brilliance/error can often define outcomes.
As such, there shouldn’t really be any reliable patterns to follow given its unpredictable nature.
But Expected Goals offers punters insight in abundance, and the tables below should aid your betting adventures no end.
First up is the current form guide according to xG principles:
Expected Goals (xG) Form Table – Last 7 Games
This shows each team’s form in their last seven matches according to the xG metric.
Obviously the data is slightly more nuanced than this, but if you want to keep it simple green indicates an excellent performance, blue a decent effort, purple a below-par showing and red something of a tonking.
The grid below shows crude xG data for home/away performances, plus an overall ‘Expected Goals Difference’. The teams are ranked by this differential:
Expected Goals (xG) League Table – Year to Date