United and xG (now that Ole is gone will things change?)

Rood

nostradamus like gloater
Scout
Joined
Jun 21, 2008
Messages
18,489
Location
@United_Hour
Used to be a fan of xG but there are now so many different numbers floating around that it makes it completely pointless - there are big descrepancies too between different algorithms (Understat, FBref, infogol, wyscout etc all have different numbers)

You'd have to take a few and work out an average to get a fair picture but who can be arsed to do that ?
 

KingCavani

New Member
Joined
Jun 27, 2021
Messages
1,264
It really bothers me how they still haven't broken down XG into situational categories. Some teams in particular will play very differently when they have a lead to defend. In Mourinho's first season our five highest XG scores came in games we drew. Antonio Conte's sides generally stand off once they get a lead while City/Liverpool just always play with the handbrake off no matter what.

People put way too much stock into as a measure of performance, it's as much a reflection on how teams play.
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
45,842
Location
Canada
It really bothers me how they still haven't broken down XG into situational categories. Some teams in particular will play very differently when they have a lead to defend. In Mourinho's first season our five highest XG scores came in games we drew. Antonio Conte's sides generally stand off once they get a lead while City/Liverpool just always play with the handbrake off no matter what.

People put way too much stock into as a measure of performance, it's as much a reflection on how teams play.
Does it matter though? All it'll say is on another day, the game that a team gad a lead in could've been drawn because they still conceded chances at the end of the day. It's not about how you play with a lead or without - if you pull back after a lead (or the other team steps up), you still have to restrict chances. You are able to split it up situationally anyway, it's just not that useful.
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
45,842
Location
Canada
Used to be a fan of xG but there are now so many different numbers floating around that it makes it completely pointless - there are big descrepancies too between different algorithms (Understat, FBref, infogol, wyscout etc all have different numbers)

You'd have to take a few and work out an average to get a fair picture but who can be arsed to do that ?
Eh, it's still just a stat but a useful stat. Nobody is using xG to decide the premier league table, it's used to help judge the quality of teams. They're all close enough to each other that the stats will say the same story.
 

Rood

nostradamus like gloater
Scout
Joined
Jun 21, 2008
Messages
18,489
Location
@United_Hour
Eh, it's still just a stat but a useful stat. Nobody is using xG to decide the premier league table, it's used to help judge the quality of teams. They're all close enough to each other that the stats will say the same story.
Well that's the thing, they are often not close at all

Understat have Palace at 0.56, Infogol at 1.04 - it's basically double

And some had us behind Arsenal on xG, others have us ahead - so a completely different picture

This kind of huge variance makes it all a bit pointless - which source are you going to use?
 

bosnian_red

Worst scout to ever exist
Joined
Aug 13, 2011
Messages
45,842
Location
Canada
Well that's the thing, they are often not close at all

Understat have Palace at 0.56, Infogol at 1.04 - it's basically double

And some had us behind Arsenal on xG, others have us ahead - so a completely different picture

This kind of huge variance makes it all a bit pointless - which source are you going to use?
Usually the differences are down to some zeroing out chances that are flagged offside but the play never stopping because nothing came of it. And besides - its very evidently not pointless. Even if there are discrepancies between sources it still gives a lot of information. Information is good. Team analysts use underlying stats like xG (whatever reliable source they use, likely something like statsbomb) to help analyze their teams. Hell even managers like Tuchel openly talk about it in press conferences. It's just a stat that strips away the bias of "shots" and goals and focuses on the quality of those chances. Because creating systems that create more good chances and preventing them is repeatable over time, while scoring low % chances is not a repeatable thing.

Public xG stats are just used to help the public analyze who is truly good and who is lucky (so far). Does it matter? I mean, nothing matters for the public because at the end of the day they are using every single stat to just talk to somebody else and say "so and so is shit" or "so and so is good". So no, it doesn't matter for the majority. It matters a hell of a lot for actual clubs, scouts, analysts, etc. But in terms of general discussion, of course they are still useful. The little discrepancies in individual games get pretty flattened out over time anyway. XG has always been a tool best used for long term analysis, while it's usable for single game analysis just on a basis of seeing how the chances were distributed. Palace being at 0.56 or 1.04 doesn't matter that much, round to the nearest whole number and that's still ~1 goal they could've scored (that huge chance off a corner, and then likely that Benteke chance that wasn't zeroed out). They still had a huge opportunity off a set piece that we were lucky to get away with, and let us go on to get the win basically.
 

Rood

nostradamus like gloater
Scout
Joined
Jun 21, 2008
Messages
18,489
Location
@United_Hour
Usually the differences are down to some zeroing out chances that are flagged offside but the play never stopping because nothing came of it. And besides - its very evidently not pointless. Even if there are discrepancies between sources it still gives a lot of information. Information is good. Team analysts use underlying stats like xG (whatever reliable source they use, likely something like statsbomb) to help analyze their teams. Hell even managers like Tuchel openly talk about it in press conferences. It's just a stat that strips away the bias of "shots" and goals and focuses on the quality of those chances. Because creating systems that create more good chances and preventing them is repeatable over time, while scoring low % chances is not a repeatable thing.

Public xG stats are just used to help the public analyze who is truly good and who is lucky (so far). Does it matter? I mean, nothing matters for the public because at the end of the day they are using every single stat to just talk to somebody else and say "so and so is shit" or "so and so is good". So no, it doesn't matter for the majority. It matters a hell of a lot for actual clubs, scouts, analysts, etc. But in terms of general discussion, of course they are still useful. The little discrepancies in individual games get pretty flattened out over time anyway. XG has always been a tool best used for long term analysis, while it's usable for single game analysis just on a basis of seeing how the chances were distributed. Palace being at 0.56 or 1.04 doesn't matter that much, round to the nearest whole number and that's still ~1 goal they could've scored (that huge chance off a corner, and then likely that Benteke chance that wasn't zeroed out). They still had a huge opportunity off a set piece that we were lucky to get away with, and let us go on to get the win basically.
Well this thread is specifically about publicly available xG stats and what they mean. I dont think most fans realise that there are many different xG numbers floating around nowadays because most stats are the same regardless of source

And to me a difference between 0.56 and 1.04 is huge - literally 100% difference and paints a completely different picture of what happened in that specific match, there are similar in every match now between the various sources

When xG first appeared you didnt see this kind of variance as there were just a couple of sources with not much difference and I used to like it as a quick snapshot of the match - but now its a total mess and you need to go and look at the specifics of what value they have placed on any specific chance which I cant be arsed with
 
Last edited:

TMDaines

Full Member
Joined
Sep 1, 2014
Messages
11,276
xG is good but you have to think of what it measures, which is just chances from shots. If a player goes one and one with the keeper, takes too long, and defender nicks it away just in time. The xG is 0 but one might say it was a great chance to score.
This.
 

Classical Mechanic

Full Member
Joined
Aug 25, 2014
Messages
34,582
Location
xG Zombie Nation
xG is good but you have to think of what it measures, which is just chances from shots. If a player goes one and one with the keeper, takes too long, and defender nicks it away just in time. The xG is 0 but one might say it was a great chance to score.
It's evolving all the time. In the future there'll be less and less blind spots. The blog below addresses some of the issues highlighted.

https://karun.in/blog/expected-threat.html
 

gajender

Full Member
Joined
May 7, 2016
Messages
2,138
Well that's the thing, they are often not close at all

Understat have Palace at 0.56, Infogol at 1.04 - it's basically double

And some had us behind Arsenal on xG, others have us ahead - so a completely different picture

This kind of huge variance makes it all a bit pointless - which source are you going to use?
Infogal has added Benteke's chance which was called offside so that accounts for the variance and In match against Arsenal both Understat and Infogal had United ahead of Arsenal on Xg ,which other source shows us behind them.
 

Classical Mechanic

Full Member
Joined
Aug 25, 2014
Messages
34,582
Location
xG Zombie Nation
Well this thread is specifically about publicly available xG stats and what they mean. I dont think most fans realise that there are many different xG numbers floating around nowadays because most stats are the same regardless of source

And to me a difference between 0.56 and 1.04 is huge - literally 100% difference and paints a completely different picture of what happened in that specific match, there are similar in every match now between the various sources

When xG first appeared you didnt see this kind of variance as there were just a couple of sources with not much difference and I used to like it as a quick snapshot of the match - but now its a total mess and you need to go and look at the specifics of what value they have placed on any specific chance which I cant be arsed with
FBref and Wyscout are the best sources that you mentioned. FBref use Statsbomb which is used by top clubs. Wyscout is also used by top clubs.
 

Rood

nostradamus like gloater
Scout
Joined
Jun 21, 2008
Messages
18,489
Location
@United_Hour
Infogal has added Benteke's chance which was called offside so that accounts for the variance and In match against Arsenal both Understat and Infogal had United ahead of Arsenal on Xg ,which other source shows us behind them.
Exactly - that is a major issue when different xG calculations are giving a completely opposite view of the match
 

gajender

Full Member
Joined
May 7, 2016
Messages
2,138
Exactly - that is a major issue when different xG calculations are giving a completely opposite view of the match
Would it possible for you to give me sources showing us Behind Arsenal in Xg because sources you have cited gave similar Xg calculation for that game.
 
Last edited:

MDFC Manager

Full Member
Joined
Dec 26, 2005
Messages
19,702
Mostuseless stat ever. It is a statistical nonsense.
I used to think the same. But when I started correlating this with the eye test, it started to make a lot of sense. It's not perfect, it's certainly acceptable though
 

Rood

nostradamus like gloater
Scout
Joined
Jun 21, 2008
Messages
18,489
Location
@United_Hour
Would it possible for you to give me sources showing us Behind Arsenal in Xg because sources you have cited gave similar Xg calculation for that game.
Wyscout: MUFC 1.92 Vs 2.22 Arse

And yes most other sources had us ahead
 

Borys

Statistics Wizard
Joined
May 10, 2013
Messages
5,800
Location
Bielsko Biala, Poland
So... A lot of talk about improved defensive game. Judging by xG parameters, we might have improved slightly on xGA since Ole left, but when you take under consideration the level of opponents we've faced in the last few games it's really not looking pretty.
Based on the conclusions I've made a year ago analyzing xG, if the opponent gets xG>0,8 it's very unlikely not to concede a goal (we average about 1).
So, the only positive aspect so far is De Gea improved form, and we're getting some momentum but we're still really shaky defensively.

It was also interesting to see we completely changed approach (=no pressing) compared to Crystal Palace game, some say it's down to fatigue, my theory is Ralf saw how little clear chances we created against Palace sitting deep and decided to invite Norwich and play them on the counter, as it suits our forwards more. Which in the end didn't make that much of a difference (as without the penalty xG would be about 1,3 vs 1,1 we got against CP), but we did have some good chances so I assume we will continue down this path rather than intensive pressing, especially after Ralf's comment about lack of intensity from forward players.
 

Chesterlestreet

Man of the crowd
Joined
Oct 19, 2012
Messages
17,395
It was also interesting to see we completely changed approach (=no pressing) compared to Crystal Palace game, some say it's down to fatigue, my theory is Ralf saw how little clear chances we created against Palace sitting deep and decided to invite Norwich and play them on the counter, as it suits our forwards more.
What do you base that on?

The stats are very comparable (pretty much identical for pressure applied in the final third).

Granted, we looked worse (more sloppy and sluggish) - but statistically, the difference seems negligible (whereas there was a considerable difference going from Carrick's last match via Arsenal to Palace).
 

Borys

Statistics Wizard
Joined
May 10, 2013
Messages
5,800
Location
Bielsko Biala, Poland
What do you base that on?

The stats are very comparable (pretty much identical for pressure applied in the final third).

Granted, we looked worse (more sloppy and sluggish) - but statistically, the difference seems negligible (whereas there was a considerable difference going from Carrick's last match via Arsenal to Palace).
My logic:
Palace had 38% possession so about 34mins, we applied 114 pressures (counting only attacking and midfield 3rd, as defensive third is not really pressing as they already pushed us into the defense). So 3,3 pressures per minute.
Then Norwich had the ball for 48% of time, which is about 43mins, we applied 101 pressures so 2,3 pressures per minute.
Difference calculated that way we pressured 40% more vs CP than vs Norwich.
We had 58 pressures in defensive third compared to 36 vs Palace, so they pushed us back a lot and the team was not so compact, what should reflect on heatmaps too.

Bare in mind Palace was shit on the ball so our pressing looked amazing, but still that's my conclusion.
 

WR10

Correctly predicted France to win World Cup 2018
Joined
Jul 19, 2009
Messages
5,626
Location
Dream
xG is good but you have to think of what it measures, which is just chances from shots. If a player goes one and one with the keeper, takes too long, and defender nicks it away just in time. The xG is 0 but one might say it was a great chance to score.
No it was defended well
 

roonster09

Hercule Poirot of the scouting world
Scout
Joined
May 10, 2009
Messages
32,049
My logic:
Palace had 38% possession so about 34mins, we applied 114 pressures (counting only attacking and midfield 3rd, as defensive third is not really pressing as they already pushed us into the defense). So 3,3 pressures per minute.
Then Norwich had the ball for 48% of time, which is about 43mins, we applied 101 pressures so 2,3 pressures per minute.
Difference calculated that way we pressured 40% more vs CP than vs Norwich.
We had 58 pressures in defensive third compared to 36 vs Palace, so they pushed us back a lot and the team was not so compact, what should reflect on heatmaps too.

Bare in mind Palace was shit on the ball so our pressing looked amazing, but still that's my conclusion.
38% possession is not 34 mins of possession. It is calculated based on number of passes.

Anyways, for Crystal palace game our PPDA was 7 which was brilliant, against Norwich it was 8.55 which isn't as good as palace but a good stat. Also yeah, in the game it looked we weren't as aggressive.
 

Borys

Statistics Wizard
Joined
May 10, 2013
Messages
5,800
Location
Bielsko Biala, Poland
38% possession is not 34 mins of possession. It is calculated based on number of passes.

Anyways, for Crystal palace game our PPDA was 7 which was brilliant, against Norwich it was 8.55 which isn't as good as palace but a good stat. Also yeah, in the game it looked we weren't as aggressive.
How exactly does that work? I'm curious as I always thought it's which team has the ball, like a binary system.
 

roonster09

Hercule Poirot of the scouting world
Scout
Joined
May 10, 2009
Messages
32,049
How exactly does that work? I'm curious as I always thought it's which team has the ball, like a binary system.
This is how Opta record possession. Total pass made by team A divided by total passes made by both teams.

Here’s how Opta describe the process now:

“Opta now record possession in a football match by means of an automated calculation based on the number of passes that a team has in a game.

We have two analysts, each monitoring one of the teams and they log each event in a game, totalling between 1600 and 2000 events per match. Each of these events has a timecode plus an x,y coordinate and the collection system is rigorously monitored by our team of checkers.

During the game, the passes for each team are totalled up and then each team’s total is divided by the game total to produce a percentage figure which shows the percentage of the game that each team has accrued in possession of the ball.”

For Opta, then, “ball possession” means the ratio of completed passes and is not a measure of time, although they claim that the two are very closely related.
 

Borys

Statistics Wizard
Joined
May 10, 2013
Messages
5,800
Location
Bielsko Biala, Poland
This is how Opta record possession. Total pass made by team A divided by total passes made by both teams.
Thanks
This bit:
For Opta, then, “ball possession” means the ratio of completed passes and is not a measure of time, although they claim that the two are very closely related.

is good enough for me to stick to the time, as it shouldn't be too far from the truth.
 

roonster09

Hercule Poirot of the scouting world
Scout
Joined
May 10, 2009
Messages
32,049
Thanks
This bit:
For Opta, then, “ball possession” means the ratio of completed passes and is not a measure of time, although they claim that the two are very closely related.

is good enough for me to stick to the time, as it shouldn't be too far from the truth.
Yeah it should be closely related, wonder how they track someone like maguire who takes 10-15 seconds to play the ball when City plays 10 passes in that time.
 

Chesterlestreet

Man of the crowd
Joined
Oct 19, 2012
Messages
17,395
My logic:
Palace had 38% possession so about 34mins, we applied 114 pressures (counting only attacking and midfield 3rd, as defensive third is not really pressing as they already pushed us into the defense). So 3,3 pressures per minute.
Then Norwich had the ball for 48% of time, which is about 43mins, we applied 101 pressures so 2,3 pressures per minute.
Difference calculated that way we pressured 40% more vs CP than vs Norwich.
We had 58 pressures in defensive third compared to 36 vs Palace, so they pushed us back a lot and the team was not so compact, what should reflect on heatmaps too.

Bare in mind Palace was shit on the ball so our pressing looked amazing, but still that's my conclusion.
Interesting - I'll have to consider this more closely.

By the way - does anyone have a good source for running stats (distance and, more importantly, sprints)?
 

roonster09

Hercule Poirot of the scouting world
Scout
Joined
May 10, 2009
Messages
32,049
Interesting - I'll have to consider this more closely.

By the way - does anyone have a good source for running stats (distance and, more importantly, sprints)?
Running stats are not available for free, usually they are not posted too except for CL games.