Andre Villas-Boas - Head Coach

[scaramanga]

Re: ***The Official AVB Discussion Thread***

*****

You edited this post before I could quote it so sorry for any paraphrasing:

You stated that a study with a general result would be an unreliable predictor for a single case - I agree with that. But the study also tells us that there are lots of other underlying issues it can't take into account. The problem is, neither can we as fans. Unless we start going into the world of conjecture and rumour, we're basing our opinions on little or no more data than the study itself.

Most of the arguments against AVB I can see can be filtered down to:

Results bad = sack the manager
and
Goals scored bad = sack the manager

In which case we're working on no more data than is present in the study. On that basis (and ignoring all the other stuff as we don't actually know anything about it), a study in points vs sacking is a perfectly valid predictor.

 

[africanspurs]

Re: ***The Official AVB Discussion Thread***

You edited this post before I could quote it so sorry for any paraphrasing:

You stated that a study with a general result would be an unreliable predictor for a single case - I agree with that. But the study also tells us that there are lots of other underlying issues it can't take into account. The problem is, neither can we as fans. Unless we start going into the world of conjecture and rumour, we're basing our opinions on little or no more data than the study itself.

Most of the arguments against AVB I can see can be filtered down to:

Results bad = sack the manager
and
Goals scored bad = sack the manager

In which case we're working on no more data than is present in the study. On that basis (and ignoring all the other stuff as we don't actually know anything about it), a study in points vs sacking is a perfectly valid predictor.

i really dont know how to respond here

i have my issues about the whole thing and some of the stuff i read directed towards people on the matter is all. I'm not trying to cause any offense here or act like i know the meaning of the universe...not at all. i just think there is scope for discussion about it

 

[BrainOfLevy]

Re: ***The Official AVB Discussion Thread***

You edited this post before I could quote it so sorry for any paraphrasing:

You stated that a study with a general result would be an unreliable predictor for a single case - I agree with that. But the study also tells us that there are lots of other underlying issues it can't take into account. The problem is, neither can we as fans. Unless we start going into the world of conjecture and rumour, we're basing our opinions on little or no more data than the study itself.

Most of the arguments against AVB I can see can be filtered down to:

Results bad = sack the manager
and
Goals scored bad = sack the manager

In which case we're working on no more data than is present in the study. On that basis (and ignoring all the other stuff as we don't actually know anything about it), a study in points vs sacking is a perfectly valid predictor.

Agreed, agreed, agreed.

Of course we don't know if in this very specific situation, it might not be better to sack AVB. But we need something else other than the above arguments to decide that we should do it.

 

[Jurgen the German]

Re: ***The Official AVB Discussion Thread***

Think a few people are unnecessarily overcomplicating things.

Surely the only "parameters" should be:

Has the manager created a good team spirit?

Does he have the full backing of the dressing room?

Are the results good enough?

Are the performances good enough and/or are there signs that things will improve?

Are the players buying into the manager's long term vision and does it fit the personnel?

Can we see the manager's vision bearing fruit even if it means some bumps in the road during the short term?

 

[africanspurs]

Re: ***The Official AVB Discussion Thread***

Think a few people are unnecessarily overcomplicating things.

Surely the only "parameters" should be:

Has the manager created a good team spirit?

Does he have the full backing of the dressing room?

Are the results good enough?

Are the performances good enough and/or are there signs that things will improve?

Are the players buying into the manager's long term vision and does it fit the personnel?

Can we see the manager's vision bearing fruit even if it means some bumps in the road during the short term?

Mate,....At least 3 of those questions are complicated :lol:

 

[africanspurs]

Re: ***The Official AVB Discussion Thread***

Agreed, agreed, agreed.

Of course we don't know if in this very specific situation, it might not be better to sack AVB. But we need something else other than the above arguments to decide that we should do it.

alright i can agree with this also

(hey, i can get used to this )

 

[Jurgen the German]

Re: ***The Official AVB Discussion Thread***

Mate,....At least 3 of those questions are complicated :lol:

Surely it's less complicated than referring to studies and theoretical analysis?

 

[billyiddo]

Re: ***The Official AVB Discussion Thread***

And to add, what i feel is the most important question - has enough time elapsed for us to be able to answer those questions with any deal of certainty?

 

[africanspurs]

Re: ***The Official AVB Discussion Thread***

Surely it's less complicated than referring to studies and theoretical analysis?

heeeeeeelllll naaw :lol:


Take this for example

"an we see the manager's vision bearing fruit even if it means some bumps in the road during the short term?"

I'll take statistical analysis over that anyday

 

[scaramanga]

Re: ***The Official AVB Discussion Thread***

i really dont know how to respond here

i have my issues about the whole thing and some of the stuff i read directed towards people on the matter is all. I'm not trying to cause any offense here or act like i know the meaning of the universe...not at all. i just think there is scope for discussion about it

I'm just trying to get some discussion going with you is all - especially as you clearly know your stuff stats wise. I'd like to hear your take on the study.

 

[africanspurs]

Re: ***The Official AVB Discussion Thread***

I'm just trying to get some discussion going with you is all - especially as you clearly know your stuff stats wise. I'd like to hear your take on the study.

aww

flattering the geek in me will get you everywhere

i'll digest the literature over lunch and get back to you. you never know, i might aactually agree with the professionals for once

 

[ringo]

Re: ***The Official AVB Discussion Thread***

Just so all the stats dunces can get the basics (I include myself, for the moment, in this group) could someone bullet point the essence of the reversion to mean argument.

In this kind of way:

1. Bunch of teams with poor results. (How many. What defines poor results)
2. Some teams sacked manager. (Any criteria)
3. Others did not.
4. There was no statistically significant difference in change in performance between those that did and those that didn't. (over what time period)

Is that it roughly?
Was this controlled for other variables?

 

[braineclipse]

Re: ***The Official AVB Discussion Thread***

Just so all the stats dunces can get the basics (I include myself, for the moment, in this group) could someone bullet point the essence of the reversion to mean argument.

In this kind of way:

1. Bunch of teams with poor results. (How many. What defines poor results)
2. Some teams sacked manager. (Any criteria)
3. Others did not.
4. There was no statistically significant difference in change in performance between those that did and those that didn't. (over what time period)

Is that it roughly?
Was this controlled for other variables?

That's roughly it I think.

Regression to the mean basically states that extreme outcomes will most likely be followed by results closer to average. This is not because there's a correction mechanism per se (7 heads in a row still leaves a 50% chance of either result on the next flip), but rather because there are more possible outcomes closer to average than at the extremes. The latter part is particularly true for a bell curve/normal distribution where the results are weighted towards the middle.

This is true for both positive and negative extremes. See the supposed manager/player of the month curse for example.

aww

flattering the geek in me will get you everywhere

i'll digest the literature over lunch and get back to you. you never know, i might aactually agree with the professionals for once

I'm guessing the biggest issue with it will be that it's not randomized as NW pointed out previously?

 

[africanspurs]

Re: ***The Official AVB Discussion Thread***

That's roughly it I think.

Regression to the mean basically states that extreme outcomes will most likely be followed by results closer to average. This is not because there's a correction mechanism per se (7 heads in a row still leaves a 50% chance of either result on the next flip), but rather because there are more possible outcomes closer to average than at the extremes. The latter part is particularly true for a bell curve/normal distribution where the results are weighted towards the middle.

This is true for both positive and negative extremes. See the supposed manager/player of the month curse for example.



I'm guessing the biggest issue with it will be that it's not randomized as NW pointed out previously?

you cant easily randomize over several leagues. each league ...much like each financial market (like we discussed about index) should have its own test BEFORE you test against each other

i wouldnt want to test continental leagues all bunched up with the premier league? the amount of things different here compared to elsewhere... no way...at least not me..we should leave that for people that get paid

for what we are checking it should be contained to the same league

infact it probably even should be contained to the same division and same financial group...the only variation before we add anything else is what we are testing....and thats sacking a dude and comparing the afters

if it was random i would be more worried about that, or rather i'd be like "i cant be arsed...dont agree. you're all WRONG!!!!""

the whole point of randomising is to eliminate bias. but in this case we actually want as much bias as possible in differing variables so that we can tell someone

if you sack someone in a the premier league from a top 6 club it will statistically make no difference in results over X length of time versus if you keep said person from the premier league in a top 6 club

 

[ringo]

Re: ***The Official AVB Discussion Thread***

That's roughly it I think.

Regression to the mean basically states that extreme outcomes will most likely be followed by results closer to average. This is not because there's a correction mechanism per se (7 heads in a row still leaves a 50% chance of either result on the next flip), but rather because there are more possible outcomes closer to average than at the extremes. The latter part is particularly true for a bell curve/normal distribution where the results are weighted towards the middle.

This is true for both positive and negative extremes. See the supposed manager/player of the month curse for example.

Even if this is the way it works, it would be the way it works "on average" presumably?

For example, Team X might sack manager A, and hire manager B who then proceeds to replace the players with 11 sheep (some might argue you couldn't tell the difference in our case). Results would then not revert/regress to mean.

Statisticians could discount this as being a post sacking managerial decision that was some distance from the average.
Fans of Team X would say what happened to that regression to the mean stuff... and call for manager B's sacking, assuming manager C would not pick sheep and results would improve due to the manager.

 

[Jordinho]

Re: ***The Official AVB Discussion Thread***

you cant easily randomize over several leagues. each league ...much like each financial market (like we discussed about index) should have its own test BEFORE you test against each other

i wouldnt want to test continental leagues all bunched up with the premier league? the amount of things different here compared to elsewhere... no way...at least not me..we should leave that for people that get paid

for what we are checking it should be contained to the same league

infact it probably even should be contained to the same division and same financial group...the only variation before we add anything else is what we are testing....and thats sacking a dude and comparing the afters

if it was random i would be more worried about that, or rather i'd be like "i cant be arsed...dont agree. you're all WRONG!!!!""

the whole point of randomising is to eliminate bias. but in this case we actually want as much bias as possible in differing variables so that we can tell someone

if you sack someone in a the premier league from a top 6 club it will statistically make no difference in results over X length of time versus if you keep said person from the premier league in a top 6 club

I agree with this. I think changes in the squad can be a massive factor. You change manager close to the end of the season and that give the players enough of a lift to improve results short term. You then go through a transfer window and suddenly it doesn't matter so much who the manager is. There has to be bigger issues than simply a run of poor results before a change is needed.

 

[scaramanga]

Re: ***The Official AVB Discussion Thread***

Even if this is the way it works, it would be the way it works "on average" presumably?

For example, Team X might sack manager A, and hire manager B who then proceeds to replace the players with 11 sheep (some might argue you couldn't tell the difference in our case). Results would then not revert/regress to mean.

Statisticians could discount this as being a post sacking managerial decision that was some distance from the average.
Fans of Team X would say what happened to that regression to the mean stuff... and call for manager B's sacking, assuming manager C would not pick sheep and results would improve due to the manager.

That's quite right, but we can only make a decision based on the facts we know.

The facts we know are

1) The performances aren't good enough

And that's it. All this "lost the dressing room"/"lacks player belief" nonsense is just that. We don't know who any future manager might be, and we don't know what they may or may not do if we put them in charge.

So on the basis that our performances aren't good enough, there are quite a few studies (listed previously) which show that changing the manager doesn't lead to any better performances than not changing (based on the tiny bit of knowledge us fans have).

That's not to say Levy should use that study as a basis, he will have far more info than us on which to base his decision, but it's good evidence that those fans who call for a manager's head purely on performance reasons are not doing so on a sound or logical basis.

 

[Mulletperm]

Re: ***The Official AVB Discussion Thread***

Please can you try and be a little more polite, there is no need for this hostility.

This made me chortle after some of the condescending veiled insults that have gone on in here recently...

 

[BrainOfLevy]

Re: ***The Official AVB Discussion Thread***

Think a few people are unnecessarily overcomplicating things.

Surely the only "parameters" should be:

Has the manager created a good team spirit?

Does he have the full backing of the dressing room?

Are the results good enough?

Are the performances good enough and/or are there signs that things will improve?

Are the players buying into the manager's long term vision and does it fit the personnel?

Can we see the manager's vision bearing fruit even if it means some bumps in the road during the short term?

1. He appeared to have done last season - see the way Bale ran to him following the West Ham winner and everyone got involved. No reason why that can't happen again.

2. Again, he appears to have done. Bad results and a particularly embarrassing will always cause question marks and a bit of descent but I can't see how after a small period of bad form players that were previously behind him will suddenly be saying 'sod it, not playing for this guy any more'.

3. Not great are they? But then again, has the sample size of results been long enough to prove that there is no way back? It hasn't. Has there been a study showing that results in a period of form like this are likely to pick up? Yes.

4. Not really, but there have been flickers this year and our problem match last season (at home to teams that are below our real level) is one we have shown better performances in this year. We never played a game like Swansea or Norwich at home last year (and I am completely acknowledging that they weren't even great performances - it shows how bad we looked last year). If there have been flickers, and considering the players are still settling in and foreign players say it takes at least 6 months to feel on some sort of good level, I think there is cause to give it time.

5. I think so. Again, not really any reason to say that they aren't. To me it clearly looks like they are trying to play the way he wants, they just aren't doing it very well yet.

6. Yes.


I can see why fans don't like concepts like regression to the mean in football. It's an emotional game and it's hard because a 5 game winless streak (for example) could see us go 2 months without remembering what it is like to see our team win a game. That's a long stretch really and it's hard to believe it will ever get better, but it almost always does. I'd say the straw that breaks the camels back is when the players to decide to jack it in with the current manager and can't see themselves that he will ever get his ideas across consistently. But Levy will know that better than anyone. And besides, I don't think it's an attitude that should be encouraged within the club. I don't want our players having the excuse that they can decide to call time on a long term project because it's tough for them right now. As if any other manager is going to come in, make sweeping changes and we all live in happy land in a magical house on gumdrop lane.

 

[ringo]

Re: ***The Official AVB Discussion Thread***

That's quite right, but we can only make a decision based on the facts we know.

The facts we know are

1) The performances aren't good enough

And that's it. All this "lost the dressing room"/"lacks player belief" nonsense is just that. We don't know who any future manager might be, and we don't know what they may or may not do if we put them in charge.

So on the basis that our performances aren't good enough, there are quite a few studies (listed previously) which show that changing the manager doesn't lead to any better performances than not changing (based on the tiny bit of knowledge us fans have).

That's not to say Levy should use that study as a basis, he will have far more info than us on which to base his decision, but it's good evidence that those fans who call for a manager's head purely on performance reasons are not doing so on a sound or logical basis.

So is the logical conclusion from that, that, on average, there is no point in changing managers to improve performance?

Of course if that were the case, every club would think that they were the exception, even though that's not possible.