Nuthin’ but an xG Thang

First they came for the Action Area stats, but I did not speak up because I wasn’t interested in Action Area stats. Next, they came for the Effective Passing Combination stats, but I did not speak up because I couldn’t care less about Effective Passing Combination stats.  Then, they came for the Heat Maps, and I did not speak up because I did not see the analytical merit in Heat Maps. Now they have come for the Expected Goals, and I haven’t the first clue what the hell they’re talking about.

Expected goals, or xG to the cool kids in the designer cagoules, is the latest statistical analysis formula to sweep the football watching nation. First used by Sam Green in a blog for Opta in 2012, xG has now found its way into our living rooms on Saturday evenings, popping up in the corner of the screen during post-match analysis on Match of the Day like Toasty from Mortal Kombat II. But expected goals seems to pose more questions than it answers – namely: What is it? What does it mean? How is it useful? And is this another sign of the liberal elite trying to brainwash the common people?

In the past decade data analysis has become de rigueur in football with professional clubs across the world employing data analysts to study player performances in training and matches, identify strengths and weaknesses of each member of the squad, and working with the coaching staff to address deficiencies within the team. Since their foundation in 2001, Opta have steadily risen to become the big dogs in sports statistics, and they enjoy strong working relationships with a host of Premier League clubs, providing in-depth analysis on team and player performance. Quickly, supporters caught onto the benefits of having a comprehensive understanding of statistics – it can give you a wider appreciation of the game, it can make you sound extremely learned down the pub, and you can also use that understanding to create great #memes. OptaJoe – the English football arm of the Opta brand – currently has 992,000 followers on Twitter, showing that there’s certainly a thirst for stats, facts and trivia in this country. So it comes as no surprise that it was one of Opta’s own that created the formula for xG.

heatmap

Expected goals measures the quality of chances created in a game by analysing a database of 300,000 shots on goal, and using this analysis to calculate how likely it is any given chance will be scored. It takes into account the position on the pitch the player finds themselves, the distance from the goal, the difficulty of the shot, and the positioning of defenders – if a player is presented with a tap-in that will be scored 9 in 10 times, it’s worth 0.9xG. Tedious arguments over who was the ‘better’ team in any given match have been given a whole new angle to cover. Before Barcelona made tiki-taka trendy, possession stats weren’t given the headline billing they have been for the past ten years. A good example of the obsession over possession came at Turf Moor last season. Burnley’s fine 2-0 victory over Liverpool was dominated by chat around the fact that they had achieved it with only 25% of the possession. The xG stats tell another story, whereby Bunley’s expected goals were 0.4 and Liverpool’s were 1.2. The takeaway being that Burnley were clinical with their half-chances, while Liverpool had a lot of the ball but did very little with it. Using this information coaches are able to draw their own conclusions – Jurgen Klopp would encourage his playmakers to work on their incisive passing in order to create chances of higher quality, while Dyche may have been happy to use the same tactical set up against the rest of the top 6 when they visited Lancashire – potentially a big part in Burnley’s excellent home form last season.

The same stats can be used in reverse to focus on the quality of a team’s defence. Even if they’re achieving clean sheets, conceding expected goals to the opposition means that aspects of their defending can be improved. Or on the flipside, a freakish 5-0 defeat whereby the opposition recorded a low number of expected goals can lead the coach to focus on other areas of improvement – probably starting by sticking someone in goal. So the idea is very simple – the higher a team’s expected goals the better the chances they create, the better at attacking they are (unless their forward is not converting them), and the more likely they are to win.

“Football is a simple game made complicated by people who should know better”
Bill Shankly

Using these stats we can look at the Premier League table after six games and get a wider picture of how teams have performed so far. Unsurprisingly, Manchester City have the highest xG in the league with 17.97 – an average of just under three goals per game. If they maintain this rate, they’ll be looking at a total xG of 113.81 by the season’s end, the current record for goals scored in a season is held by Chelsea at 104. City also have the lowest xG against at 3.42. Having scored 21 and conceded 2 they’ve made the most of their chances, but also been fortunate not to concede more. The highest xG against is, shock horror, Crystal Palace, with 12.87 – almost bang on the thirteen they’ve actually conceded. What is a genuine surprise is that Palace aren’t bottom for expected goals for, having racked up an xG of 6.92 in six games so far, but scoring diddly squat. Brighton and Hove Albion currently have the lowest xG with only 3.15 clear cut chances created so far – their goal tally of five showing that they’ve been able to convert difficult chances. The biggest disparity between xG for and against is at Bournemouth, whose expected points total of 3.5 puts them bottom of the (expected) league. Having only mustered an xG of 3.55, their leaky defence has conceded 10.63 xG – it might be an idea for Eddie Howe to have a word with Nathan Ake and tell him to stop bombing forward. The worst converters in the Premier League, after Crystal Palace of course, are Newcastle, whose xG of 9.51 makes their actual tally of 6 look a little woeful – but might be put down to the mercy Joselu showed Stoke at St James Park. Chelsea are the most efficient with their chances, converting 12 of their 6.76 expected goals.

In a season where much of the debate has surrounded the various strikers in the Premier League, the emergence of xG is a perfect storm. While there is no measure greater than scoring goals, drilling into the data allows us to analyse the effectiveness of leading strikers. The likes of Sergio Aguero and Gabriel Jesus are odds-on to challenge for the Golden Boot given the amount of chances their teammates create, but are they as clinical – and therefore as valuable – as their competitors? Using the data from the opening six games and a sample of players that have scored 3+ goals, Alvaro Morata is currently the most effective forward in the league. His six goals in six games have arrived from an xG of 3.41 – meaning at least two of his goals have arrived from positions where he wouldn’t be backed to score. Morata’s closest competitor is Raheem Sterling, who has started the season in a rich vein of form. Five goals in five games with an xG of 3.36 is an excellent return from a player that isn’t utilised in a central position. Given his traditionally slow start to the season, its not surprising to see that Harry Kane has the lowest conversion to xG rate, with his four goals falling just short of the expected return of 4.48. Romelu Lukaku has been the best provided striker in the league with an xG of 6.38 – fans of throwing money at bookmakers could do a lot worse than having a flutter on the Belgian for top scorer. Light can also be shed on the quality of chances provided and Sergio Aguero leads the table for xG assists, laying on a potential 3.62 goals so far this season – down in no small part to his partnership with Jesus. David Silva and Henrikh Mkhitaryan, the league’s top two assist makers, are just behind the Argentinian, while Newcastle’s Matt Ritchie sits in fourth, having created opportunities for 2.3 xG.

So how much does xG tell us about the game? Well, depending on your opinion it can give us a whole host of new information or it can tell us nothing at all. It’s a beacon of hope for fans whose team are struggling – Crystal Palace supporters can look at their xG and travel with games in the knowledge that they’ll at least get a chance. Conversely, Chelsea fans might look at their team’s xG and use it as a stick to beat Conte with – obviously that would be mental but you can’t put anything past them. Or you could just ignore it completely.

Statistics in football are there for those that want them. If you want to pour over the pass completion statistics of every midfielder in the Veikkausliiga, then good luck to you. You’re the kind of person xG is for - a number cruncher, the thinking football fan, the owner of a 1993 Malawi home shirt. If you think it’s a load of old bollocks and you’d rather skull ten pints of Carling and shout abuse whenever Mark Hughes appears on your television then that’s okay too.

After all, the league table doesn’t lie – unless it does.

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