Welcome to the world of Twenty3 xG
It wasn’t that long ago that Expected Goals (xG) was the footballing equivalent of Lord Voldemort; few dared mention the phrase out loud. Some people what xG would do to the game, others feared how they’d be viewed for using it.
Now, though, we see xG popping up on Sky Sports and BBC’s Match of the Day. It is now generally accepted as part of the pool of headline stats, alongside possession, passes, shots, corners and, of course, goals.
However, there’s still much to come from xG. Here at Twenty3, thanks to our new advanced metrics, we now have a plethora of xG-related stats to help with analysis and to provide deeper insight.
Firstly, there is xG overperformance. For this total, expected goals are taken away from actual goals. This gives a better idea of which players are running hot and which are unlucky right now.
For example, this season – at the time of writing – Heung-min Son ranks first for xG overperformance across Europe’s top five leagues. The Spurs No.7 has exceeded his xG total by 5.7 goals having racked up an xG total of 2.3. Erling Haaland ranks second for overperformance having scored 4.11 more goals than he was expected to have given the chances he’s had.
On the opposite end of that, we have Kevin de Bruyne. The Manchester City maestro missed a penalty against Liverpool and this goes some way to explaining why he’s currently underperforming his expected goals total by 3.52. Only one other player across the top five European leagues is underperforming by more than two goals. Everton’s Richarlison has struck just once this term – again, at the time of writing – despite posting an xG total of 3.3.
To really delve deeper into why these players are underperforming, there’s post-shot expected goals (PSxG). This combines the quality of the shot with the quality of the effort. If a player’s PSxG is below their xG then they’re devaluing their chances with their shot placement. Shots in central areas tend to be of poorer quality and thus get a lower PSxG score.
De Bruyne’s PSxG total of 2.84 is significantly down on his xG of 5.52. As you can see in his shot placement map (above), only 11 of the City man’s 42 shots have hit the target. His radar has been a little off this season, especially in open-play given both of his goals have come via the spot.
Richarlison is having similar struggles. The Brazilian has converted just one of his 27 shots this season and despite hitting the target with 48% of his efforts. However, of those shots, 54% of them have hit the bottom third of the goal, right in the centre.
By comparison, over half of Son’s efforts in the Premier League this season have hit wide areas, not central ones.
More impressively, perhaps, is that Son has hit the target with 13 of his 22 shots this season. He’s hitting the target and aiming for the corners. As a result, he’s scoring goals. It’s why his PSxG total is 1.1 greater than his xG total. What we can take from this is that this level of overperformance isn’t sustainable, but he’ll likely outperform his xG due to his shot placement.
Haaland is unique in the sense his PSxG numbers don’t exceed his xG averages, yet he’s forever outperforming the latter metric. One explanation for this could be that he takes his shots ridiculously early. Keepers aren’t set so the efforts don’t have to hit the corners and therefore don’t rank highly on either xG or PSxG models. For now…
There’s also the capability to filter by left and right foot xG. There are some players who don’t really have a weaker foot so the “show them onto their weaker foot” shout is redundant.
The most ambipedal player in Europe’s top five, at least according to xG, is Roma’s Henrikh Mkhitaryan. The former Manchester United attacker has an xG total of 5.47 this season and the split between right and left is almost 50%. You can also use this filter to identify which players pose very little threat on their weaker foot and use historical data to see who has worked on improving their foot usage.
The possibilities are endless and the scope for analysis is massive with Twenty3’s suite of xG metrics.
All the graphics and visualisations in this article use Wyscout data and were produced in the Twenty3 Toolbox.
If you’d like to learn more about the new Twenty3 Advanced Metrics, or our other products or services, and how they might be able to help you, don’t hesitate to get in touch.