Tottenham: How the xG data foresaw Kane’s rise to stardom

Manchester City are leading,“He should have scored from there”. “That was an easy chance”. “Those misses are going to come back to haunt us”. “Yet another chance goes begging”.

These are some of the stuff that we hear fans and pundits say when a player has failed to find the back of the net from a considerably great position.

However, recently pundits, football nerds and fans have increasingly started to use a new data – by the name of Expected Goals or xG – to support such claims.


What is Expected Goal or xG?

I first came across this word while going through in-depth analysis articles and videos.

Like everybody else, it confused me as well. Of course, it is not as easy as understanding something like ‘shots on target‘ or ‘possession percentage’. It is something more complex.

Being not a big fan of intricate numbers and formulas, I initially gave xG a swipe to the left. Soon I realised that it was being used by more and more people and hence I was forced to learn more about it just like I was forced to assimilate Pythogorean Theorem.

However unlike the mathematic theory, xG actually ignited some kind of interest in me probably because it is related to football.

This interest led to me to read several articles about the same, watch videos and listen to podcasts. I got so determined to learn about this new favourite analytic tool of football data nerds that I went to read, watch or listen some of them more than once.

After all that research, if one is to ask what is Expected Goals or xG to me I would dumbstruck for more than a few seconds.

xG is “a measure of chance quality” in the words of Duncan Alexander, the co-founder of one of the most popular football data providers Opta. In the most simplest of words, it is an analytic tool that helps us to determine the quality of a goalscoring chance.


The birth and growth of xG

While I referred Expected Goals to as ‘a new data’, it has been around for about five years. It was introduced by Opta after being inspired by analytical models in the American sports.

Several others worked on Opta’s model and made changes to it, hence resulting the existence of various numbers of xG models. However, all the models give similar numbers.

Football clubs started using Expected Goals to analyse team performances and soon it made its way into media.

BBC’s Match of the Day has been extensively using xG in their shows since the start of the season. However, BBC’s use of the data has attracted quite a lot of criticism as they are using it out of context according to many.

Apart from media, the fans also have started using xG to understand their club’s or their players’ performances better.


How is the xG calculated?

“Opta have analysed over 300,000 shots to calculate the likelihood of a shot being scored from a particular position on the pitch during a particular phase of play. The model takes into account several variables and looks at how these affect the chance of a specific shot going in”, Duncan says.

The ‘several variables’ include assist type, which part of body is been used to make the shot, distance from goal, angle of the shot, etc. Some models even take into account the positions of the opposition defenders (Opta’s model does not).

The closer the shot is taken from the goal, the higher the xG. The narrower the angle of the shot, the lower the xG. If the shot is a header, the xG is lower. And so on.

Every shot is given an expected goals value (after analysing the aforementioned 300,000 shots) which is the same as the percentage chance of a particular shot ending up in the back of the net.

Expected goals value is always between 0 and 1. One is the highest xG value but it is almost impossible for a shot to have it. For example, if there is a 75% chance of a shot resulting in a goal the xG of that shot is 0.75.

A team’s (or player’s) expected goals tell us how many goals they (or he) should have scored from their chances.


What more do xG tell us?

If a striker is consistenly scoring more goals than his xG value, then he is considered to be an elite striker.

A player’s xG in a particular game or in a number of games is not what we mean by ‘consistent’ here. We are talking seasons.

Harry Kane is someone who was labelled as a one season wonder by many in his breakout season, but the Englishman has gone on to become one of the best strikers in the world.

According to, Kane’s xG in the 2014/15 season was 17.16. However he actually scored 21, about four more than his xG. The next season he had an xG value of 22.73, when he actually scored 25 goals.

Last season Kane scored 29 goals to run away with the Golden Boot, but his xG was just 19.82. In the current season, in which also he is the top scorer, the player’s goal scoring tally of 24 is better than his xG of 23.42.

I put in a little bit more effort to find out Kane’s xG before the 2014/15 season and to my surprise I was actually able to find sometjong interesting.

In the 2013/14 season, Kane played a total of 403 minutes from which he rattled the net thrice and it was almost equal to his xG of 3.2. It means that the Spurs striker showed signs of staredom even before he broke into limelight.

The metric was not as popular in 2014 and not many would have bothered to check the xG of a 20-year-old striker playing for an underperforming Spurs, but when you look back it is indeed fascinating. Today, clubs will be closelt looking at the xG values of a ‘next Messi’ or ‘next Ronaldo’ along with other data before signing.

This is how Expected Goals works with players and strikers. It helps us understand how good or bad a player is performing in terms of goals and being clinical.

With regards to how Expected Goals works with teams, one prominent example that I came across was of Juventus’s 2015/16 season.

Juve had won only three of their first ten games of the season. However, their xG showed how good they were actually performing. They scored 11 goals and conceded nine during those ten games, but their expected goald were 19 and the expected goals of their opponents were just five.

So, the xG metric showed that the Bianconeri’s results were not as good as their performances suggested.

Soon they started scoring more than their xG value and conceding less than their opponents’ xG before going on to clinch the title with a nine-point-lead over the second placed Napoli.



Expected Goals do not solve football or predict results, but when used properly it help us understand the game better.

In a game team A might have a better xG than team B but the result would have gone 1-0 in favour of team B. It simply means that the winners were more clinical.

Anyone who watched the game would be able to tell that, but Expected Goals is a number that confirms it. It also confirms or denies our claims of ‘he should have scored that’.

When used in the right context, there is a lot that Expected Goals can do and that is why its admirers see it as the new real deal in football analytics.

P.S: If you do have any doubts or questions on xG, please free to drop in a comment. We would be happy to get back to you guys.
as they march on to win.


Written by Dakir Thanveer

Follow Dakir on Twitter @ZakWriter

Like O-Posts on Facebook

You can also follow O-Posts on Twitter @OPosts