Xavi Hernández

[delancey]

Exactly. This is a fucking second tier European competition and yet we can't even win it or come close. And than 5hat idiot Xavi talks about getting pissed of hearing CL anthem because we aren't playing in it lol.

In the round of 32 no less

 

[Givenchy]

Fireable offence to start Alonso over Christensen as a CB in this game. No one in the right frame of mind would ever consider Alonso as a better choice. Some European success would do wonders for the confidence of the squad, but Xavi is just self sabotaging. Ended up having to fix his mistake by subbing on Christensen and Balde anyway.

Not only that, he subbed off our only threat in Raphinha or was he resting him for Cadiz?

Literally braindead

 

[Cool]

With Pedri and Gavi out for the Man United return leg I think we should revert back to our 433 with Busi as the lone partnered by Kessie and De Jong.

Along with that, we should cross our fingers and hope to god that Busi rolls back the years and the defense tightens themselves up because I get shivers imaging Busi trying to defend all the space left behind with Rashford and co attacking us in old trafford.

 

[delancey]

With Pedri and Gavi out for the Man United return leg I think we should revert back to our 433 with Busi as the lone partnered by Kessie and De Jong.

With a false 9?

 

[delancey]

Not only that, he subbed off our only threat in Raphinha or was he resting him for Cadiz?

Literally braindead

Haha I like your sense of humor. I think Raphinha is always they guy who gets substituted regardless of how well he plays.

 

[Cool]

With a false 9?

No, Lewa as our 9 flanked by Dembele on the RW and Torres/Fati, or Dembele on the LW and Raphinha on the RW.

 

[gregorrin10]

Just play the best fucking defense in the return leg. Play fucking Roberto-Garcia-Alonso-Alba complete joke/meme of a defense against Cadiz for all I care. And give TORRE a fucking game there too! And if he impresses, then also put him in the squad for the return leg. With both Gavi and Pedri likely out there's gonna be room/a need in the squad anyway, and in any case he'd be useful to have as a sub for either midfield or LW for that game too. He could grab this chance with both hands, but he needs to get that chance first!

If I see more than one player of the BACK defense start against Cadiz, I'm probably gonna be betting big on United going through, cause you just know he'll then play the same fucking shit he did today.

 

[Cool]

knowing xavi he?s probably gonna play alonso as leftback in the return leg 😂💀💀

 

[Birdy]

Your assertion is there is no difference in value between off an on target shots. What you then proceed to do is provide an edge case scenario. Would you agree, that in expectation (on average), on target shots are more valuable than off target shots. Because the vast majority of off-target shots aren't "ones skyrocketed from in front of an empty net".

Notice how this relates to my above point. Too arrogant, not critical thinking enough.

Let's talk seriously then:
The area you are touching is related to all the Xstatistics. The fundamental principle of calculating xG (which is the stat that measures by definition the value of the shot) is that it doesn't take into consideration where the shot ended.
Factors that most models use in their calculations are: - locations of the shot, distance from goal, players in between the ball and the GK, positioning of the GK, speed of the ball the moment of shot, positioning of the body of the shooter, etc. And many others, the models are becoming increasingly sophisticated.
The statistical methodological assumptions you mention are all in there, and vary from model to model. Of course, no model is perfect.

The expectations of a shot becoming a goal has to do with all the above, which are all PRIOR to the action. The end result of a shot, on the contrary, occurs AFTER the action. This means that it doesn't yield 'expectation'...
Now, what you are possibly trying to insinuate is that the end result of a shot can indicate a backwards correlation with scoring a goal, a 'reverse' expectation so to speak. There might be a correlation there, that's why people thought in the past the on/off target stat is important and were reporting it.
However, that would be substantially weaker than the correlation between G and xG, which is forward-oriented correlation. It looks at anterior conditions that statistically lead to a goal, instead of trying to statistically hypothesize from posterior conditions. Just as with any kind of backwards causality, backwards correlation is in a way putting the cart before the horse.

Data science in football has moved the last 10 years towards xStats for such reasons, and totally away from looking at end result stats (goals, assists, wins, points, etc), let alone stats that can only vaguely be backwards-correlated to the target end results stats (like on/off target or possesion), which was the norm 40 years back for instance.

PS: If you are actually doing data science in your real life, it is surprising that you don't take the xStats more seriously than you seem to take them in all the threads here. People working in football data science, are employing the same tools that you do in a different field

PS2: It's funny that many of you accuse me of arrogance, because I hold and defend my views. Usually I have an argument for that, I don't do it because of any fetish. And usually I do not change views easily, because the argument I have seems to be stronger than opposing arguments.
On the contrary, I see the exact same people accusing me of arrogance being super arrogant in their own views, even when these views are less well-founded and less backed by arguments, but more backed by consensus of the majority.

 

[gregorrin10]

How the fuck can he start with that defence?

Was pointed out how ETH would target Alonso with Rashford which he did.

I'm pretty sure that as soon as he saw Alonso in the lineup, ETH literally re-shifted his whole attacking line to the point of playing Weghorst as a fucking CAM (and Bruno totally out of position too) just so that he could put Rashford against Alonso. Yet Xavi kept playing Araujo on the right. It's like he was brain dead at that point, I swear. At least change Kounde and Araujo around then, for fuck's sake. And Rashford clearly knew and was instructed to stick more to the right to really target Alonso and Alba specifically. And that's exactly where both their goals came from.

 

[delancey]

No, Lewa as our 9 flanked by Dembele on the RW and Torres/Fati, or Dembele on the LW and Raphinha on the RW.

Dembele is injured

 

[Cool]

Dembele is injured

He wont be fit for the return leg?

 

[delancey]

He wont be fit for the return leg?

Nope, it?s already next week

 

[KingLeo10]

Let's talk seriously then:
The area you are touching is related to all the Xstatistics. The fundamental principle of calculating xG (which is the stat that measures by definition the value of the shot) is that it doesn't take into consideration where the shot ended.
Factors that most models use in their calculations are: - locations of the shot, distance from goal, players in between the ball and the GK, positioning of the GK, speed of the ball the moment of shot, positioning of the body of the shooter, etc. And many others, the models are becoming increasingly sophisticated.
The statistical methodological assumptions you mention are all in there, and vary from model to model. Of course, no model is perfect.

The expectations of a shot becoming a goal has to do with all the above, which are all PRIOR to the action. The end result of a shot, on the contrary, occurs AFTER the action. This means that it doesn't yield 'expectation'...
Now, what you are possibly trying to insinuate is that the end result of a shot can indicate a backwards correlation with scoring a goal, a 'reverse' expectation so to speak. There might be a correlation there, that's why people thought in the past the on/off target stat is important and were reporting it.
However, that would be substantially weaker than the correlation between G and xG, which is forward-oriented correlation. It looks at anterior conditions that statistically lead to a goal, instead of trying to statistically hypothesize from posterior conditions. Just as with any kind of backwards causality, backwards correlation is in a way putting the cart before the horse.

Data science in football has moved the last 10 years towards xStats for such reasons, and totally away from looking at end result stats (goals, assists, wins, points, etc), let alone stats that can only vaguely be backwards-correlated to the target end results stats (like on/off target or possesion), which was the norm 40 years back for instance.

PS: If you are actually doing data science in your real life, it is surprising that you don't take the xStats more seriously than you seem to take them in all the threads here. People working in football data science, are employing the same tools that you do in a different field

PS2: It's funny that many of you accuse me of arrogance, because I hold and defend my views. Usually I have an argument for that, I don't do it because of any fetish. And usually I do not change views easily, because the argument I have seems to be stronger than opposing arguments.
On the contrary, I see the exact same people accusing me of arrogance being super arrogant in their own views, even when these views are less well-founded and less backed by arguments, but more backed by consensus of the majority.

What I, and many others, have been trying to communicate to you goes a lot deeper than "no model is perfect". It's not just that the model isn't perfect, it's HOW FAR from perfect it is:

" locations of the shot, distance from goal, players in between the ball and the GK, positioning of the GK, speed of the ball the moment of shot, positioning of the body of the shooter, etc. And many others, the models are becoming increasingly sophisticated."

Notice how none of these parameters (as far as you've written here) capture: 1) Who is taking the shot (are they GOAT tier, clutch under pressure etc.) ; 2) Who is the GK (are they prime Neuer or are they MATS in CL KOs); 3) are one or both teams on a hot streak or losing streak, among countless other factors that common sense tells dictate high leverage situations/ games.

When I say the methodology needs to be questioned more, what I mean is that the relatively simple (used precisely because they are easy to quantify) parameters do not capture the essence of high leverage football games. The model is a very simplified/poor representation of reality and thus cannot be trusted wholly. Sure, we can gain some trends/insights for further investigation but at this point in time, I find it hard to label these insights as conclusive as you do.

Ultimately, the success of any model is based on its predictive power. Without running any sort of statistical models, I can tell you a team with Messi/Pep on it will dominate a league in terms of trophies and a team named RM will dominate CL in terms of results. Teams are remembered both for style of play and results. Ideally, you achieve both (prime Barca, prime Milan), but if it's between results (not just Ws but actual big trophies like CL, LL, EPL) and style of play, I'd weigh the former more.

 

[iniestaGOAT]

I'm pretty sure that as soon as he saw Alonso in the lineup, ETH literally re-shifted his whole attacking line to the point of playing Weghorst as a fucking CAM (and Bruno totally out of position too) just so that he could put Rashford against Alonso. Yet Xavi kept playing Araujo on the right. It's like he was brain dead at that point, I swear. At least change Kounde and Araujo around then, for fuck's sake. And Rashford clearly knew and was instructed to stick more to the right to really target Alonso and Alba specifically. And that's exactly where both their goals came from.

Hahaha and they still couldnt beat us.

Its going to be very different at old taffiord my son.