Exhausted by xG?

An economics professor at my university once stated in a lecture that “economic forecasts are only right half of the time, so we may as well just flip a coin!” Whilst this was a throw away comment meant not to be taken particularly seriously, the comparison between complex statistical models used to measure probability and pure chance is a compelling one and one that I often think back to when I see statistics used in such a definitive way as they have often been to analyse Albion’s Premier League woes this season.

Albion’s season is one where the narrative has become overwhelmed by one particular complex statistical model. xG, A model used to show the expected goals scored by each team, but one that has often not been matched by the end result in Albion matches this season. There have been multiple games that xG predicted that an Albion victory was most probable, but which wasn’t achieved. Mostly because of so many missed goalscoring opportunities and a fair amount of sloppy goals given away at the other end.

This was most dramatically highlighted in Albion’s home defeat to rivals Crystal Palace who beat Albion 2-1 after managing just 3 shots, 2 of which were on target, compared to Brighton’s 25 shots yielding just one goal. That this defeat was then followed up with a defeat to West Brom where Brighton missed 2 penalties and a host of other guilt edged chances after their opponents scored early on from a corner, one of only two shots they managed on target, only increased the attention on Brighton’s misfiring strikers.

Furthermore, Brighton have had an xG advantage over their opponents of more than 1 expected goal 7 times this season, but have only beaten their fellow relegation strugglers Newcastle (twice) by a winning margin of more than one goal. The numbers tell a clear story, Albion have underperformed in front of goal this season.

We could pull these kinds of stats out about many more of Brighton’s games this season till we are blue in the face. But I’ll stop there because you’ve probably read this kind of stuff a number of times before. And yet it keeps getting discussed, mainly because it’s an easy explanation as to why Brighton have looked so good during much of the season, yet this isn’t reflected in the league table

However, despite Albion’s lack of clinical edge in front of goal, goalscoring isn’t actually a massive issue relative to its nearest rivals. After 29 games plays Brighton had the second best goal difference, had scored the third most, conceded the least and lost the least times of all the teams in the bottom half of the Premier League table.

But whilst these stats highlight some positives and improvements in Albion’s play, it is not being seen in an improvement in points gained and league position. Despite xG suggesting Albion have been better than they have managed to achieve in terms of results, they remain in their common Premier League standing just above the league’s bottom three. So the consistent and dramatic deficiency when compared to XG is raising some eyebrows, but possibly points to a deficiencies in the model itself.

We could go into detail about the ins and out of the model itself, but I don’t know how much value that would have. But in general I think that Albion are showing up the flaws of placing a lot of merit in a weighted probability model such as xG to measure the likelihood of goals being scored in a low scoring game such as football. And to a degree the weight we place on statistics generally in football.

We can all get pulled in by data analysis at times, especially when they argue for a point that we believe to be true. The increase in the use of statistics and analytics has helped us to understand the game in ways we previously didn’t. But they don’t often tell you the full story of an event, or the full reasons as for why a team dominates a particular game or a particular period of play. Maybe an early goal, an in-game injury, a number of minor tactical shifts, physical conditions or the connotations a certain result has on their league position, or simply individual errors makes a team change its approach. There are many things these generic weighted probability models simply don’t account for.

For Albion, their accumulated xG performance after 29 games puts them better off than Man United and West Ham and 4th in the xG table. In reality they are 22 points and 12 league places worse off. However only five of the 20 Premier League teams sit in their xG projected position, with 9 more than 2 positions different than this projection. Including Everton, who according to xG would sit 17th rather than their existing 8th position.

So it’s not just Albion’s whose variation compared to xG is far from accurate. But many have taken these statistical anomalies and run with them. Some have said that simply if Albion signed a £40/50m striker then they would start taking those chances and qualify for Europe, which is a huge simplification. Especially considering Albion have thrown away 17 points from winning positions this season. Albion’s problems go much further than simply not scoring enough goals.

In reality Brighton have constraints in terms of their aspirations. In particular its wage bill, which has consistently been one of the lowest in the league during its time in the topflight and yet they have spent most of their time at the top level out of the relegation zone and are not far from securing a club record fifth consecutive topflight season. Focusing on individual statistics like xG can at times mask the bigger picture. One which shows in reality Albion and Everton are probably about where they should expect to be rather than at opposite ends of the table.

Whatever the opinion, statistics don’t tell you how to solve a problem, at best they tell you how it was created. And that’s only if they data has been collected and analysed accurately and is then interpreted correctly. Much of which leads into areas of judgement and subjectivity rather than fact.

The solutions for these problems the data highlights are created by the experts. The creative minds like Graham Potter and his regular tactical switches. Minds that most football fans deep down wish they could pit their wits against, but in reality are greatly inferior to in this context.

Football is a game of opinions, not statistics. Moments of ingenuity and creativity, moments that aren’t created by data analysis of historic points in time alone, but by the invention and astuteness of those at the very top of their profession in the moment.

Whatever the data does show, the fact that Albion are sitting six points and two places above the Premier League relegation zone in their fourth consecutive topflight season should be considered an overperformance rather than an underperformance. We should stop focusing on xG and focus more on what a remarkable period of Albion history that we are currently watching.

Author: tweetingseagull

A Fan of Brighton and Hove Albion and all things Football. Follow my tweets here: https://mobile.twitter.com/TweetingSeagull

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s