Expected Goals Table

[Tomo]

A friend just sent me this. I know little about it but he is a massive stat man....

Cringeworthy reading the table but tells the story!

 

[Carey 6]

Not sure it does tell much of the story at all really. We’ve been extremely low on this for a while. 

We are doing well when it comes to taking chances, we’re just not creating many. We’re one of the more high scoring teams in the championship I believe. 

The defence & amount of goals conceded is what’s letting us down. 

[JoeAman08]

15 minutes ago, Tomo said:

A friend just sent me this. I know little about it but he is a massive stat man....

Cringeworthy reading the table but tells the story!

 

Been talking about this table for weeks. Some call it nonsense but I disagree. We have been in the bottom 3 of that table for months. The chances were we were always going to drop well out of the playoff picture.
We don’t create many chances but earlier in the year we were finishing almost all the chances we had. Meanwhile Bentley has kept quite a few out that would normally be goals. Lately our goals have dried up to an extent but Bentley is as heroic as before. Also, Eliasson has accounted for a lot of points on his own as well. Honestly, think we have been lucky to get 35 points on the board so quickly so should stay out of a relegation battle but since Afobe was injured our performances have been some of the worst I can remember in my short time going up to AG

[cidercity1987]

The only thing I would question about expected goals is how on earth is it calculated? I lost a little respect for the model when I saw Diedhiou's goal against Fulham, dead centre tap in from a lovely rolling square ball with no-one in sight, was only 50% chance of a goal being scored.

https://experimental361.com/2019/12/08/championship-timelines-6-8-dec-2019/ Exhibit one, Diedhiou v Fulham approx. 50% chance.

https://experimental361.com/2019/12/27/championship-timelines-26-dec-2019/ Exhibit two, Charlton's third yesterday almost 100% chance of a goal, but with the goalie to beat.

That's a nonsense.

[JoeAman08]

11 minutes ago, cidercity1987 said:

The only thing I would question about expected goals is how on earth is it calculated? I lost a little respect for the model when I saw Diedhiou's goal against Fulham, dead centre tap in from a lovely rolling square ball with no-one in sight, was only 50% chance of a goal being scored.

https://experimental361.com/2019/12/08/championship-timelines-6-8-dec-2019/ Exhibit one, Diedhiou v Fulham approx. 50% chance.

https://experimental361.com/2019/12/27/championship-timelines-26-dec-2019/ Exhibit two, Charlton's third yesterday almost 100% chance of a goal, but with the goalie to beat.

That's a nonsense.

Some would say they have seen Fam miss those 

Joking aside, that is good info. Cheers. Does just only account for the goal scoring chance or the build up? Only asking because we were only really probing then burst into life and 3-4 passes later we scored. 

[And Its Smith]

xG is absolute nonsense. The whole thing doesn’t work one bit 

[Alex_BCFC]

Only person who creates chances really is Elliasson. We ain't going to score many goals with him on the bench.

[Zookeeper WSM]

12 minutes ago, RedDave said:

xG is absolute nonsense. The whole thing doesn’t work one bit 

I think it pretty accurately reflects the table which shows that it’s a strong predictive model but highlights some interesting outliers like us, stoke and West Brom which tells a story. I’ve been following it closely all season and it’s been a pretty accurate reflection for me. It basically tells us that we’ve only been the stronger team in two matches which is exactly how I’ve seen the season. Bentley has been superb and kept us in games and we’ve converted a good proportion of our chances. Being bottom three in expected goals and in the play off zone in the real table is unsustainable and we’re starting to see it play through. 

[And Its Smith]

2 minutes ago, Zookeeper WSM said:

I think it pretty accurately reflects the table which shows that it’s a strong predictive model but highlights some interesting outliers like us, stoke and West Brom which tells a story. I’ve been following it closely all season and it’s been a pretty accurate reflection for me. It basically tells us that we’ve only been the stronger team in two matches which is exactly how I’ve seen the season. Bentley has been superb and kept us in games and we’ve converted a good proportion of our chances. Being bottom three in expected goals and in the play off zone in the real table is unsustainable and we’re starting to see it play through. 

Couldn’t disagree more. It’s a ridiculous model that cannot accurately define how good a chance is thus making the whole thing wrong 

[Ron W]

22 minutes ago, RedDave said:

xG is absolute nonsense. The whole thing doesn’t work one bit 

That's the thing, it's not perfect but it's not nonsense. Do we create a lot of clear-cut chances? No, and it tells that story. It's worrying that we are the worst team in the division for it. Some people were saying on here a month or so ago that we can't rely on Bentley to save us every week or someone to pop up with a goal out of nothing and they're being proved right at the moment.

[Coxy27]

35 minutes ago, RedDave said:

xG is absolute nonsense. The whole thing doesn’t work one bit 

I suggest betting syndicates who use it as an important measure would disagree with you, and have large pots of cash as evidence.

[Lrrr]

1 hour ago, JoeAman08 said:

Some would say they have seen Fam miss those 

Joking aside, that is good info. Cheers. Does just only account for the goal scoring chance or the build up? Only asking because we were only really probing then burst into life and 3-4 passes later we scored. 

Only the shot itself 

[And Its Smith]

32 minutes ago, Coxy27 said:

I suggest betting syndicates who use it as an important measure would disagree with you, and have large pots of cash as evidence.

xG said we beat Blackburn quite comfortably. Nonsense I say. 

[054123]

We’ve scored 36 goals this year, we scored 2 away from home yesterday!

We are the exception to the XG algorithm as we take the majority of our good chances.

Its the goals against that is the problem

[Lrrr]

2 minutes ago, 054123 said:

We are the exception to the XG algorithm

Well, Stoke as well

[054123]

Just now, hodge said:

Well, Stoke as well

The numbers are growing! 

[Coxy27]

2 minutes ago, RedDave said:

xG said we beat Blackburn quite comfortably. Nonsense I say. 

The xG in the Blackburn game was a draw (it was 1.3 v 1.2 - may very slightly depending on what company you use as the measure).

I think that is fair, Williams had a very good chance, Diedhiou did too. We had a good spell in the first half, but got done twice through our own poor play. I don't think there was anything in the overall pattern of the game to suggest the two sides were unbalanced.

It isn't designed to measure individual matches really, the real aim is for it to give an indication of the medium term fortunes of a club.

For a while now it would have indicated we were over-performing, and I personally have thought from watching us this season that results have in many cases been better than performances. Four results isn't conclusive proof of anything, but perhaps we are reverting to the mean a little now. Time will tell.

[Zookeeper WSM]

9 minutes ago, RedDave said:

xG said we beat Blackburn quite comfortably. Nonsense I say. 

It didn’t, it said it was a very even game. Anything where the two teams’ expected goals are within 33% is an expected draw. Obviously we lost as as Blackburn were more clinical in that game. No one will claim expected goals is going to tell you will / should have won every game. It is a very good way to show who created / gave away the best chances in a game which over time broadly reflects where teams will be in the table and highlights outlier teams that are either very clinical / very wasteful or have ridden their luck. The key things that would be called out from the table as it stands would be that we will finish lower than we currently are if we continue to play the way we have been all things being equal and stoke are likely to finish higher.

[And Its Smith]

14 minutes ago, Zookeeper WSM said:

It didn’t, it said it was a very even game. Anything where the two teams’ expected goals are within 33% is an expected draw. Obviously we lost as as Blackburn were more clinical in that game. No one will claim expected goals is going to tell you will / should have won every game. It is a very good way to show who created / gave away the best chances in a game which over time broadly reflects where teams will be in the table and highlights outlier teams that are either very clinical / very wasteful or have ridden their luck. The key things that would be called out from the table as it stands would be that we will finish lower than we currently are if we continue to play the way we have been all things being equal and stoke are likely to finish higher.

1.24 v 0.46 

16 minutes ago, Coxy27 said:

The xG in the Blackburn game was a draw (it was 1.3 v 1.2 - may very slightly depending on what company you use as the measure).

I think that is fair, Williams had a very good chance, Diedhiou did too. We had a good spell in the first half, but got done twice through our own poor play. I don't think there was anything in the overall pattern of the game to suggest the two sides were unbalanced.

It isn't designed to measure individual matches really, the real aim is for it to give an indication of the medium term fortunes of a club.

For a while now it would have indicated we were over-performing, and I personally have thought from watching us this season that results have in many cases been better than performances. Four results isn't conclusive proof of anything, but perhaps we are reverting to the mean a little now. Time will tell.

1.24 v 0.46

[Coxy27]

9 minutes ago, RedDave said:

1.24 v 0.46 

1.24 v 0.46

I don't know where you are getting that from, but it is incorrect.

Also - you're deliberately ignoring the point 'Zookeeper WSM' and I made.

[And Its Smith]

15 minutes ago, Coxy27 said:

I don't know where you are getting that from, but it is incorrect.

Also - you're deliberately ignoring the point 'Zookeeper WSM' and I made.

If there are different xG’s available then it shows what a nonsense it is 

[Northern Red]

xG had Spurs beating Bayern Munich earlier this season, whereas the actual football went and spoilt everything by having Bayern win 7-2.

It's bollocks for people who want football to be played on spreadsheets.

[And Its Smith]

53 minutes ago, Coxy27 said:

The xG in the Blackburn game was a draw (it was 1.3 v 1.2 - may very slightly depending on what company you use as the measure).

I think that is fair, Williams had a very good chance, Diedhiou did too. We had a good spell in the first half, but got done twice through our own poor play. I don't think there was anything in the overall pattern of the game to suggest the two sides were unbalanced.

It isn't designed to measure individual matches really, the real aim is for it to give an indication of the medium term fortunes of a club.

For a while now it would have indicated we were over-performing, and I personally have thought from watching us this season that results have in many cases been better than performances. Four results isn't conclusive proof of anything, but perhaps we are reverting to the mean a little now. Time will tell.

xG has constantly had Stoke as performing better than their results. Their results haven’t changed. 

Fans are ever more desperate for an opinion of things they haven’t seen or don’t know about and xG fulfils that need 

[Coxy27]

13 minutes ago, RedDave said:

xG has constantly had Stoke as performing better than their results. Their results haven’t changed. 

Fans are ever more desperate for an opinion of things they haven’t seen or don’t know about and xG fulfils that need 

You really should research before you post...

Stoke's first 10 games - 8 losses and two draws - 2 points from 30 - 7% of available points taken.

Second 10 games - 4 wins and 6 losses - 12 points from 30 - 40% of available points taken.

Their next 4 games - 2 wins, a draw and a loss - 7 points from 12 - 58% of available points taken.

A major improvement as the season has gone on - they're not in the relegation zone now which is remarkable given their start.

[Davefevs]

xG is not nonsense, but it ain’t perfect either.  It’s based on a big sample of chances.

Even if you take 10 one on ones that are say worth 0.5.  There’s nothing to define that if you have two of them (1.0), that you score one of them.

That is the common misunderstanding and a 1 match timeline is not representative.

I use them to see the ebb and flow of chances throughout a game and the quality of those chances, but rarely say we should’ve won or lost a game based on xG.  I might say we were a bit unlucky, but as we all know the only thing that really matters is actual goals!

[And Its Smith]

41 minutes ago, Coxy27 said:

You really should research before you post...

Stoke's first 10 games - 8 losses and two draws - 2 points from 30 - 7% of available points taken.

Second 10 games - 4 wins and 6 losses - 12 points from 30 - 40% of available points taken.

Their next 4 games - 2 wins, a draw and a loss - 7 points from 12 - 58% of available points taken.

A major improvement as the season has gone on - they're not in the relegation zone now which is remarkable given their start.

Wonder when they will reach where xG says they should be!! 

[Ron W]

1 hour ago, Northern Red said:

xG had Spurs beating Bayern Munich earlier this season, whereas the actual football went and spoilt everything by having Bayern win 7-2.

It's bollocks for people who want football to be played on spreadsheets.

The idea of it is to give an indication of the quality of chances being created. It's not giving a definitive answer as to who should win, it's not meant to. After the actual Spurs-Bayern match most people were in agreement everything Bayern hit went in.

It makes a lot more sense than using shot count as a mark of performance, which we've all been guilty of.

[Coxy27]

11 minutes ago, RedDave said:

Wonder when they will reach where xG says they should be!! 

You're missing the point.

You've got an agenda against it for whatever reason and I can see I won't change your mind.

Happy holidays.

[And Its Smith]

3 minutes ago, Coxy27 said:

You're missing the point.

You've got an agenda against it for whatever reason and I can see I won't change your mind.

Happy holidays.

You’re just using it to fulfil your agenda I reckon so let’s agree to disagree

Merry Xmas 

[Mr Popodopolous]

3 hours ago, cidercity1987 said:

The only thing I would question about expected goals is how on earth is it calculated? I lost a little respect for the model when I saw Diedhiou's goal against Fulham, dead centre tap in from a lovely rolling square ball with no-one in sight, was only 50% chance of a goal being scored.

https://experimental361.com/2019/12/08/championship-timelines-6-8-dec-2019/ Exhibit one, Diedhiou v Fulham approx. 50% chance.

https://experimental361.com/2019/12/27/championship-timelines-26-dec-2019/ Exhibit two, Charlton's third yesterday almost 100% chance of a goal, but with the goalie to beat.

That's a nonsense.

I believe that is more reflective on how many goals they would be expected to score based on their all round game etc, as opposed to merely the chance. Happy to look again at it though! In other words, Diedhiou with that tap in outperformed his expected output by 0.5. Doesn't work so well in individual games but over a season, some sites show it for individual players- could be interesting. Understat (dunno if it uses the same model), shows Top Scorers in various Leagues and compares their 'real' vs their anticipated output. Could be useful as a predictor of near and medium term future returns.