Statistics in Football (Expected Goals (xG)

[CEA_20]

Welcome to another version of the Modern Fitba newsletter! With international week here (again), there will be no match previews this week, but we will give you a close look at the current Expected Goal numbers for the league as a whole.

Before we do, you should also check out Jamie Kilday’s article on Lawrence Shankland and whether the stats of the newest member of the Scotland team lives up to the current hype around him.

Expecting You

Eight games are a decent sample for looking at xG and trying to predict patterns for the rest of the season, but there is still likely to be a large degree of statistical noise to shift through before you can try to make any sort of conclusions. In terms of comparing teams within a league the main cause of such noise after 8 games is likely to be the fixture list. A team’s xG might not look impressive right now, but if they have an overload of tough opponents at the start of the season we could reasonably expect them to improve. For every week of games played we’ll gain more information and especially after the next three rounds: when all teams have faced each other once we can slowly start drawing out such likely future patterns with more certainty. 

At this stage, it’ll be beneficial to look at the xG numbers in three different ways. The first is the overall xG Table

  

This table shows the difference between the quality of chances a team has created (xG for) and the quality of chances they have conceded (xG against). This looks beyond the results of each game and indicates the quality of the actual performances, through what they actually created and prevented at each end of the pitch.

Rangers have opened up a gap in the table down to Celtic and is now 4.8 xG in front of them: a sizable gap. There is of course some differences in the two top clubs’ fixture list so far, with Celtic having played five away games to Rangers’ three and the champions also having to go to Ibrox and Fir Park – the home of the two other teams in the top 3 of the xG standings. After round 6 the gap between the two team was only 0.1, with Rangers creating this current gap through big hammerings of Aberdeen and Hamilton at home, as Celtic struggled away to Hibs and Livingston.

So do such single-game blowouts (Ranger’s 4.54 xG game difference over Dons is the largest we’ve ever recorded) skew the table somewhat? As a sense-check of this – especially at this early stage of the season – it can be useful to allocate the points in each game based on the xG difference. In the table below, the match results have been determined by xG, with a difference of less than 0.5 xG resulting in a draw (excludes penalties). Rangers’ 2-point league lead is still intact; a sign that both they and Celtic have pretty much got the results they’ve 'deserved’ based on their performances so far. It’ll be interesting to see the overall picture once the next three games have been played and all Premiership teams have faced each other. 

One team that have suffered in the xG table due to some heavy defeats is Hibs. Overall they have the second worst xG difference in the league at -4.9, but if game points were allocated purely based on xG, they would be 6th: Paul Heckingbottom’s team have had some heavy xG defeats by Rangers, Celtic and Motherwell, while in other games they’ve been equal or just better than their opponent xG wise – but they’ve not managed to transform that into points on the board to a sufficient degree.

At the bottom of the league table we find St. Johnstone. And it’s quite difficult to understand why. Their overall xG difference is 5th and if they were allocated points based on individual game xG performances, they would be 3rd!

Which leads us on to our final xG angle: the over/under performing table. This ranks the teams based on the largest difference between their actual goals difference and their expected goals difference.

St. Johnstone have by far the largest negative difference, i.e. their goal difference (excluding pens and own goals) is over 12 goals worse than it ‘should be’, based on the quality of the chances they’ve created and conceded. There is a huge gap down to Hibs, who has a negative difference of just over 5.

It’s also interesting to note that while being 4th in the league - only 1 point above 7th and 11 behind 1st - is probably not the start Aberdeen would have ideally wanted, they should consider themselves fairly lucky: Modern Fitba have them as the third worst team in the league xG wise, and they are only one spot better off with points allocated per xG. 

The xG table after 8 games is a snapshot: a better indicator of the true relationship between the team than the actual results and goasl, but not yet comprehensive enough to predict too much with certainty about the rest of the season. The Modern Fitba team have worked on and developed different kind of predictive models for the Premiership season. We’ll be talking more about these as the season progresses, but for now you can have a look at Jason’s Projected League Table and Seth’s Project Points and Team Ratings  

We’ll be back next week with a full preview of every single Premiership game!

[garydavidson]

Shouldn't you take @RandomGuy out for dinner before you start talking dirty to him?

[RandomGuy]

[Radford 72]

Thanks for posting that. I think it could have stayed in Saints Chat as it's very relevant to us currently. I think it possibly explains why there isn't the mass panic that you'd generally associate with a team without a win. 

Our play certainly looks like it's merited a lot more. 

[cristo]

My head is on fire

[south inch]

They do say that there are lies, damned lies and statistics!  The only table that counts is the one that shows us in 12th place with 4 points.  I have not started to panic (yet) simply because I think we have good players and a good manager and years of watching Saints tells me that the tide will turn.  Only hope it does so bloody soon!

[dunblanemike]

 Believe Radford has been out-radforded!