The Week in Stats: Hope for Huddersfield and Madrid
By Warren Pegg
This week we look at non-shots-based expected goals in the EPL and La Liga, and see why Spurs, Huddersfield, Real Madrid and Atleti all stand out from the pack.
One thing that sets the excellent FiveThirtyEight website apart from other publicly accessible xG models is that it offers stats for non-shots-based xG, in addition to an orthodox xG model.
Non-shots-based xG (NSxG) calculations look at the attacking positions that a side has managed to occupy during a game and then, by comparing them to thousands of other very similar situations in previous matches, deduce how likely these openings were to result in a goal.
FiveThirtyEight itself describes the model as “an estimate of how many goals a team could have scored given their non-shooting actions in and around their opponent’s penalty area”.
The biggest benefit here is that NSxG stats pick up on important events that are completely disregarded by orthodox shots-based xG models. For instance, let’s say that a move ends with a player needing to complete a simple pass inside the six-yard box in order to create an open-goal scoring opportunity. However, the player in question then fails to complete that simple pass, and the ball rolls harmlessly out of play.
As no shot took place, a traditional xG model won’t register this sequence of play at all, despite the fact that it would often have resulted in a goal being scored. In contrast, NSxG stats do register such events.
So comparing a team’s shots-based xG and NSxG numbers allows us to see which sides are failing to capitalize upon the promising openings that they’re creating.
There are limitations to what this can tell us, of course. While we can agree that not converting high NSxG positions into shots or goals is a bad thing, explaining precisely why that might be happening is much harder to assess.
For example, it could be down to bad luck or the manager imposing a flawed attacking style that makes it unusually difficult to convert possession in dangerous areas into shots.
Alternatively, it could mean that the team’s creative players are performing poorly when it comes to making the final pass, or that a side’s forwards are failing to adopt good positions at decisive moments.
Similarly, a team’s over-achievement in terms of its NSxG numbers might be due to good fortune, the exceptional individual talent of certain players, or an idiosyncratic attacking style whose effectiveness the model doesn’t pick up on.
Despite these limitations, though, comparing xG and NSxG figures can still allow us to pick out important broad themes and identify issues worthy of closer investigation.
Huddersfield, with only 12 goals to their name in the Premier League so far, are by some margin the EPL’s worst under-achievers in terms of converting their openings into attempts on goal. But there may be some cheering news here for fans of the West Yorkshire club, or at least reason for them to be less pessimistic.
Because although Huddersfield have created by far the lowest-quality chances in the league in terms of shots-based xG – a total of 15.6, with the next worst being Newcastle’s 19.9 – in NSxG terms they’re creating better openings than four of their relegation rivals: Cardiff, Burnley, Newcastle and Brighton.
Meanwhile, during an uneven and unpredictable season, Spurs have over-achieved in NSxG terms far more than any other Premier League club. In the absence of publicly available NSxG data for previous seasons, it’s hard to tell whether this performance is likely to be sustainable or revert to the mean, but it’s certainly something to keep an eye on as the campaign progresses.
Finally, moving on to look at La Liga, the two Madrid giants have been the worst culprits when it comes to not capitalizing upon the openings they create. This goes some way to explaining why, despite Real Madrid having finished every season this decade as one of La Liga’s two highest-scoring clubs, this term not just Barcelona but also Sevilla, Levante and Celta Vigo have all netted more goals than Los Merengues.