Val Ball and stats
#31
(02-03-2022, 09:38 AM)Sliced Wrote:
(02-03-2022, 09:29 AM)Derek Hardballs Wrote: How are the stats are created? Who / what judges what a good, half chance is? If it’s a person then that is subjective rather than fact.

Sorry, do you think someone sitting in a room is deciding what xG each shot is?

No I was asking a question and software is only as good as the person coding it. Doesn’t mean it’s bullet proof as a method to determine whether a team has played well.
Reply
#32
I tell you what he didn't do and that is analyse the stats during the game. He just sat back, watched us lose, then went back to the lab.

" I don't understand it Igor, the stats say we won the game yet we got no points, what am I missing Igor?"
" I don't know master, maybe we should just try something different"
" Never Igor, this is the plan, the one and only plan and it will work, it must work, and when it does........ we will rule the World Ahhhhh, hahahaha"
Igor under his breath "Fucking lunatic"
Reply
#33
This is agenda in stats is such tedium now. It’s a balance.

There’s a place for pragmatism, guy feeling and statistics. It’s not one or the other.

Stats have a significant part to play, but if the stats are incongruous with results and performance then something has to change and it didn’t with Val. xG is just one stat in a collection that should be used as a whole.

Ok common misconception is xG doesn’t account for the quality don’t he chance, it does. It also backs off the type of chance, against historical data of that chance or changes if that type being taken. So it is a fair reflection of whether you are performing well in front of goal.

However, for 10 or so games our xG has been bloody woeful so no idea why Val thought we were still creating quality chances.
Reply
#34
A few observations from the thread (Sorry I can't be bothered to quote and reply)...

XG isn't a stat for "interpretation" its a robust assessment of the quality of chance v the quality of finishing. It is blunt and to the point. Based on thousands and thousands of previous data entries.

XG doesn't tell you if you "played well" as that is a subjective topic based on your personal preferences. It tells you whether or not you created good enough opportunities, bluntly.

Stats in general are a way of finding out where the problems in performance are factually, and help us remove bias and prejudice. But as football is emotive, it is easier to deal in bias and prejudice as its easier to understand.

Using statistics to underpin all decisions in football is the right thing to do. But you also have to allow the subjective to play a part as well as be governed by preference at some level.

So instead of debate g the use of stats just as a away of lamenting Val, yet again, we should have a poll about what "brand" and philosophy we all want. As sorry to say stats ay going anywhere and will always underpin decisions.
"IMO of course..."
Reply
#35
(02-03-2022, 10:56 AM)BurleyBaggie Wrote: This is agenda in stats is such tedium now. It’s a balance.

There’s a place for pragmatism, guy feeling and statistics. It’s not one or the other.

Stats have a significant part to play, but if the stats are incongruous with results and performance then something has to change and it didn’t with Val. xG is just one stat in a collection that should be used as a whole.

Ok common misconception is xG doesn’t account for the quality don’t he chance, it does. It also backs off the type of chance, against historical data of that chance or changes if that type being taken. So it is a fair reflection of whether you are performing well in front of goal. 

However, for 10 or so games our xG has been bloody woeful so no idea why Val thought we were still creating quality chances.

How many times do Mass and I have to state we aren’t saying it’s either or. How come those who are so pro-stat seem incapable of reading what was written by those who have said it has its place but only as one tool at the managers disposal. Val let stats get in the way of the glaringly obvious, and that was it wasn’t working and for a sustained period of time. At that point the only stat he should have been concerned with was the points accrued.
Reply
#36
(02-03-2022, 10:42 AM)Derek Hardballs Wrote:
(02-03-2022, 09:38 AM)Sliced Wrote:
(02-03-2022, 09:29 AM)Derek Hardballs Wrote: How are the stats are created? Who / what judges what a good, half chance is? If it’s a person then that is subjective rather than fact.

Sorry, do you think someone sitting in a room is deciding what xG each shot is?

No I was asking a question and software is only as good as the person coding it. Doesn’t mean it’s bullet proof as a method to determine whether a team has played well.

It's not to determine whether a team has player well, it's an indicator of how good are the chances that they've had in a game.

xG takes a database of millions of shots and looks at historical chances that are from the same position/same speed of pass prior to the chance/whether its a header or shot as the current chance and sees how often chances that are identical or as close to identical as possible resulted in a goal (broadly speaking, there's some smoothing involved too but that will change the result by 0.001 type magnitudes).

The point is it's based upon whether shots have gone in in the past in identical/similar situations. What judges whether it is a good chance or not is what has happened over millions of shots in the past. It's as watertight as it can possibly be, and it's very simple coding that can be verified by the most junior of data scientists with the same data.

(02-03-2022, 09:55 AM)MassDebater Wrote:
In what World did Hugill's stats get him a job
, unless it was purely how far he runs in games?! Seriously. What stats does he have that mean he should be playing anywhere? Oliver Burke is another prime example. Lots of mediocre players keep getting massive deals, but their stats surely would not warrant it. Then again that's why I write code and am not a massively rich football manager. It's the one job where you can be pretty shit and still get more highly paid work.

Well it's certainly not his xG stats is it? Maybe people were using their eyes instead and coming to the wrong conclusion that he could be a productive goal scoring striker
Reply
#37
(02-03-2022, 11:15 AM)Sliced Wrote:
(02-03-2022, 10:42 AM)Derek Hardballs Wrote:
(02-03-2022, 09:38 AM)Sliced Wrote:
(02-03-2022, 09:29 AM)Derek Hardballs Wrote: How are the stats are created? Who / what judges what a good, half chance is? If it’s a person then that is subjective rather than fact.

Sorry, do you think someone sitting in a room is deciding what xG each shot is?

No I was asking a question and software is only as good as the person coding it. Doesn’t mean it’s bullet proof as a method to determine whether a team has played well.

It's not to determine whether a team has player well, it's an indicator of how good are the chances that they've had in a game.

xG takes a database of millions of shots and looks at historical chances that are from the same position/same speed of pass prior to the chance/whether its a header or shot as the current chance and sees how often chances that are identical or as close to identical as possible resulted in a goal (broadly speaking, there's some smoothing involved too but that will change the result by 0.001 type magnitudes).

The point is it's based upon whether shots have gone in in the past in identical/similar situations. What judges whether it is a good chance or not is what has happened over millions of shots in the past. It's as watertight as it can possibly be, and it's very simple coding that can be verified by the most junior of data scientists with the same data.

(02-03-2022, 09:55 AM)MassDebater Wrote:
In what World did Hugill's stats get him a job
, unless it was purely how far he runs in games?! Seriously. What stats does he have that mean he should be playing anywhere? Oliver Burke is another prime example. Lots of mediocre players keep getting massive deals, but their stats surely would not warrant it. Then again that's why I write code and am not a massively rich football manager. It's the one job where you can be pretty shit and still get more highly paid work.

Well it's certainly not his xG stats is it? Maybe people were using their eyes instead and coming to the wrong conclusion that he could be a productive goal scoring striker

Well we shall see if those same attacking players create and convert more chances now we have a change of manager won’t we.
Reply
#38
(02-03-2022, 11:10 AM)Derek Hardballs Wrote:
(02-03-2022, 10:56 AM)BurleyBaggie Wrote: This is agenda in stats is such tedium now. It’s a balance.

There’s a place for pragmatism, guy feeling and statistics. It’s not one or the other.

Stats have a significant part to play, but if the stats are incongruous with results and performance then something has to change and it didn’t with Val. xG is just one stat in a collection that should be used as a whole.

Ok common misconception is xG doesn’t account for the quality don’t he chance, it does. It also backs off the type of chance, against historical data of that chance or changes if that type being taken. So it is a fair reflection of whether you are performing well in front of goal. 

However, for 10 or so games our xG has been bloody woeful so no idea why Val thought we were still creating quality chances.

How many times do Mass and I have to state we aren’t saying it’s either or. How come those who are so pro-stat seem incapable of reading what was written by those who have said it has its place but only as one tool at the managers disposal. Val let stats get in the way of the glaringly obvious, and that was it wasn’t working and for a sustained period of time. At that point the only stat he should have been concerned with was the points accrued.

Well you obviously don't read, comprehend or do research as you're opening post is factually incorrect hence my ire:

the stats said we were creating more chances than the opposition but that only tells a fraction of the story, most of those chances weren’t gilt edged they were half chances from poor crosses

That is simply not how this one stat works - it is predicated on looking at the quality of the chance and the historical data.

The issue isn't the stats, its the fact that they weren't used properly and that things like xG (as per my post) were dire in his last 10 games. There was no way he could have been using that stat to defend his position.
Reply
#39
(02-03-2022, 11:29 AM)Derek Hardballs Wrote:
(02-03-2022, 11:15 AM)Sliced Wrote:
(02-03-2022, 10:42 AM)Derek Hardballs Wrote:
(02-03-2022, 09:38 AM)Sliced Wrote:
(02-03-2022, 09:29 AM)Derek Hardballs Wrote: How are the stats are created? Who / what judges what a good, half chance is? If it’s a person then that is subjective rather than fact.

Sorry, do you think someone sitting in a room is deciding what xG each shot is?

No I was asking a question and software is only as good as the person coding it. Doesn’t mean it’s bullet proof as a method to determine whether a team has played well.

It's not to determine whether a team has player well, it's an indicator of how good are the chances that they've had in a game.

xG takes a database of millions of shots and looks at historical chances that are from the same position/same speed of pass prior to the chance/whether its a header or shot as the current chance and sees how often chances that are identical or as close to identical as possible resulted in a goal (broadly speaking, there's some smoothing involved too but that will change the result by 0.001 type magnitudes).

The point is it's based upon whether shots have gone in in the past in identical/similar situations. What judges whether it is a good chance or not is what has happened over millions of shots in the past. It's as watertight as it can possibly be, and it's very simple coding that can be verified by the most junior of data scientists with the same data.

(02-03-2022, 09:55 AM)MassDebater Wrote:
In what World did Hugill's stats get him a job
, unless it was purely how far he runs in games?! Seriously. What stats does he have that mean he should be playing anywhere? Oliver Burke is another prime example. Lots of mediocre players keep getting massive deals, but their stats surely would not warrant it. Then again that's why I write code and am not a massively rich football manager. It's the one job where you can be pretty shit and still get more highly paid work.

Well it's certainly not his xG stats is it? Maybe people were using their eyes instead and coming to the wrong conclusion that he could be a productive goal scoring striker

Well we shall see if those same attacking players create and convert more chances now we have a change of manager won’t we.

Who said anything about the manager?

We could convert every single shot we have from now until the end of the season and it wouldn't change the fact that we were failing to convert chances earlier in the season?
Reply
#40
(02-03-2022, 08:35 AM)SophLad Wrote: I read it as another post deriding the over-use of stats in the game.  Personally; I think stats should be used to validate the opinions of experts, not shape them.

This.

Contrary to what Derek may believe I don't tend to look at any stats after a game.

But when there is a debate about why we score so few goals, I might look at some stats to see if the opinion I have formed from watching games is justified.

Ans maybe, just maybe, if someone had looked at the playing records of some of our recruits prior to them joining us, we might not have such a goal shy squad.
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)