Christian Benteke

[scaramanga]

I am afraid you are just trying to adapt a statistical theory to footballers form. What I am telling you is that it is not appropriate to do so. You need to make so many assumptions regarding so many variables as to make it a meaningless excersise in this context. For example, can you tell me what a players 'mean' form is? You may be able to retrospectively show a correlation for a regression to mean but you simply have no way of being able to hypothecate that a players very best year will be his last at that level until after his career is over. Some players improve year on year. That is why you simply can't say that Benteke or anyone else will never be as good again as their best year. Hence the examples I gave you.

I'm not sure why you persist with this. Please don't take this the wrong way, but trying to discuss something about which you clearly have such little (or, I think, no) understanding doesn't do you any favours.

I'm not sure how much more simply I can phrase it, so I'll go with the gravity analogy again - saying "you can't apply statistical theory to football" is like saying "you can't apply gravity to football".

Clearly you can.

Finding the best season isn't really relevant, finding outliers is. Benteke's first season in this country was a clear outlier. That doesn't mean his second season won't be worse, but when we're dealing with the unknown we can only play the odds. The odds were overwhelmingly in favour of his second season being worse.

In all likelihood next season could be better than this one buy until he regularly puts in seasons like his first it's far more likely than not to be a one-off. Unless we know for certain that it's more than a one-off he's absolutely not worth the £30M that Villa were asking for. I'm not even sure he's any better than Kane unless he can regularly have seasons like his first over here.

 

[scaramanga]

Here you are pirate, an explanation in far simpler terms than I'm capable of using:

http://plus.maths.org/content/maths-minute-regression-mean

 

[Finney Is Back]

I'm not sure why you persist with this. Please don't take this the wrong way, but trying to discuss something about which you clearly have such little (or, I think, no) understanding doesn't do you any favours.

I'm not sure how much more simply I can phrase it, so I'll go with the gravity analogy again - saying "you can't apply statistical theory to football" is like saying "you can't apply gravity to football".

Clearly you can.

Finding the best season isn't really relevant, finding outliers is. Benteke's first season in this country was a clear outlier. That doesn't mean his second season won't be worse, but when we're dealing with the unknown we can only play the odds. The odds were overwhelmingly in favour of his second season being worse.

In all likelihood next season could be better than this one buy until he regularly puts in seasons like his first it's far more likely than not to be a one-off. Unless we know for certain that it's more than a one-off he's absolutely not worth the £30M that Villa were asking for. I'm not even sure he's any better than Kane unless he can regularly have seasons like his first over here.

I guess the question is whether or not it was 'clearly an outlier'. The season before he'd got 19 in 37 for Genk.... So in 2013 Benteke's following season at Villa could've looked like an even better season following a very good one.

It's difficult for us because every player that we sign is likely to be a player who has had an 'outlier' of a season. If they have a great season and then an even better season immediately after then they move into the desires of the mega rich and out of our price bracket. It means we either have to buy on potential alone with no evidence, take a punt on a player either during or immediately after their first really proper good season (like we did on Lamela). Or take a punt on a player who has had a great season and then suffered a dip in form or injury or something like that and then hope that the player can rediscover that earlier form.

One thing I will say though is that if signing a player for a significant amount of money then I would rather we signed a player who had shown that they could produce good form in England, even if only for a single season.

To link all this back to Benteke. It would be interesting to see what a club would be prepared to spend on him now..... We could've had him for £25 million in 2013 after his excellent first season in the PL. I think he would probably still go for £18M or so now. So effectively his price was a £7 million inflation off the back of that excellent season - not a bad premium to pay really (after all it was possible that he MIGHT do it again immediately after or do something even better?).

 

[scaramanga]

It's difficult because every player that we sign is likely to be a player who has had an 'outlier' of a season. If they have a great season and then an even better season immediately after then they move into the desires of the mega rich and out of our price bracket. It means we have to take a punt either during or immediately after their first really good good season (like we did on Lamela). Or take a punt on a player who has had a great season and then sufferred a dip in form or injury or something like that and then hope that the player can rediscover that form.

One thing I will say though is that if signing a player for a significant amount of money then I would rather we signed a player who had shown that they could produce good form in England, even if only for a single season.

To link all this back to Benteke. It would be interesting to see what a club would be prepared to spend on him now..... We could've had him for £25 million in 2013 after his excellent season. I think he would probably still go for about £18 million or so now. So effectively a £7 million inflation off the back of that excellent season - not a bad premium to pay really (after all it was possible that he MIGHT do it again).

That's just far too much money for one decent season though.

Where we've succeeded the most is with the likes of Modric/Berbatov/Eriksen (good performance in a poor league that doesn't command the fees) or Bale/Bentaleb where they've had no good seasons but they're so young and talented we can work with them.

We're not City/Chelsea - we can't afford PL premiums for what might be one season wonders. Maybe when we're in the new stadium and regularly in the CL, but right now that's not the right risk profile for a club in our position.

Edit:
Any reliable source had Villa saying they wouldn't even listen to offers under £30M that season. I agree that at £15M with some add-ons then he's at the right value but I still think there are other options at least as good.

 

[Finney Is Back]

Those sources were clearly not actually reliable then scara.

I do agree with just about everything else you have written though (although I wouldn't really call the German league poor)

 

[scaramanga]

Those sources were clearly not actually reliable then scara.

Again, please don't feel insulted but IMO they're more reliable than a bloke on a message board

 

[Pirate55]

Nope, if you compare his current performance to his past he's likely to regress upwards.This is a disproportionately bad season for him.

There you go making assumptions again as to what is his "mean".

Football is not an exact science. To try and "predict" whether a player is going to get better or worse than his previous "best" season the following one is shown by numerous examples to be flawed. You are not doing yourself any favors by trying to apply a statistical theory to football. If you could do this, you could make a fortune offering your service to football clubs as a consultant as to who they should buy. Clearly you can't. And neither can anyone else - otherwise this "science" would currently be being applied just as training, fitness and diet science is now being applied.

 

[Pirate55]

So we should be extremely worried about Lamela then?

Excellent point.

We should never have bought Soldado on the regression theory basis

 

[scaramanga]

There you go making assumptions again as to what is his "mean".

Football is not an exact science. To try and "predict" whether a player is going to get better or worse than his previous "best" season the following one is shown by numerous examples to be flawed. You are not doing yourself any favors by trying to apply a statistical theory to football. If you could do this, you could make a fortune offering your service to football clubs as a consultant as to who they should buy. Clearly you can't. And neither can anyone else - otherwise this "science" would currently be being applied just as training, fitness and diet science is now being applied.

Clubs do employ statistical consultants - they spend a ****ing fortune on them. Not sure where you've been for the past decade but that's been quite common for some time now.

Can we just agree that you're not willing/able to understand regression?

It's not what your post shows you clearly think it is. Try reading the link I sent you and start there. Once you understand the concept then maybe we can have a discussion about it. Otherwise I'm not sure what else I can do to help someone who can't/won't be educated.

 

[Pirate55]

Clubs do employ statistical consultants - they spend a ****ing fortune on them. Not sure where you've been for the past decade but that's been quite common for some time now.

Can we just agree that you're not willing/able to understand regression?

It's not what your post shows you clearly think it is. Try reading the link I sent you and start there. Once you understand the concept then maybe we can have a discussion about it. Otherwise I'm not sure what else I can do to help someone who can't/won't be educated.

I agree. We are just going round in circles. I am afraid I wont be persuaded that this type of analysis has any relevance to determining whether a player is going to improve or regress.

You said earlier that Benteke may or may not improve ans that we must wait until he has put performances in his standout season on a regular basis. However, the point is that if we wait to see if he does, we certainly wont be able to afford him!

According to your theory, several questions for you:

Will Bony score more or less goals in 2015?

Should Lampard ever have been allowed to leave Chelsea?

Will Defoe be a hit at Sunderland?

Will Ade ever play like he once did?

Will Townsend ever come good?

 

[milo]

I agree. We are just going round in circles. I am afraid I wont be persuaded that this type of analysis has any relevance to determining whether a player is going to improve or regress.

You said earlier that Benteke may or may not improve ans that we must wait until he has put performances in his standout season on a regular basis. However, the point is that if we wait to see if he does, we certainly wont be able to afford him!

Rather than argue against the principle of regression to mean (which is pretty much universally accepted), I think that you would be better trying to demonstrate that his bad seasons are an outlier rather than his good ones.

 

[Pirate55]

Rather than argue against the principle of regression to mean (which is pretty much universally accepted), I think that you would be better trying to demonstrate that his bad seasons are an outlier rather than his good ones.

I am not arguing against the principle of regression to mean. I am throwing doubt on its applicability as a footballer forceasting tool.

Please answer this, will Kane "regress to mean"? If so, please explain what we can expect from him in the following seasons.

Has Bale "regressed to mean", or Modric, or Messi or Ronaldo? Surely all have had long enough careers for you to be able to say. What about Lampard?

 

[milo]

I am not arguing against the principle of regression to mean. I am throwing doubt on its applicability as a footballer forceasting tool.

Please answer this, will Kane "regress to mean"? If so, please explain what we can expect from him in the following seasons.

Has Bale "regressed to mean", or Modric, or Messi or Ronaldo? Surely all have had long enough careers for you to be able to say. What about Lampard?

I don't think that Scara or anyone else has said that regression to mean can be used to forecast, we are talking about probability not prediction.

Obviously with Kane the sample size is too small to say.

With regards to very good players and regression, you might find this answers that better than I can (which goes back the curse of Sports Illustrated point I made earlier)

http://whywereason.com/2011/06/25/the-myth-of-the-si-jinx/

 

[Pirate55]

I don't think that Scara or anyone else has said that regression to mean can be used to forecast, we are talking about probability not prediction.

Obviously with Kane the sample size is too small to say.

With regards to very good players and regression, you might find this answers that better than I can (which goes back the curse of Sports Illustrated point I made earlier)

http://whywereason.com/2011/06/25/the-myth-of-the-si-jinx/

As the article concludes "you can have a lucky day, but you cant have a lucky career". That is why the regression to mean theory cannot be used as a forecast only as a retrospective tool. Great players continue to improve throughout their careers. At the outset, no-one can know what their "mean" is. That is the point.

 

[billyiddo]

id say it's probably something taken in to huge consideration when assessing the risk v reward of any potential signing.

seems like common sense really : is this player on an upward curve or has he hit his peak?

 

[milo]

As the article concludes "you can have a lucky day, but you cant have a lucky career". That is why the regression to mean theory cannot be used as a forecast only as a retrospective tool. Great players continue to improve throughout their careers. At the outset, no-one can know what their "mean" is. That is the point.

I think that you are either misunderstanding what you are reading or cherry picking the bits that you think supports what you are saying.

It is possible that Benteke could go on and become a great player. I have not seen anyone argue otherwise. What is being argued about is how likely is this? One way that we can get an idea of how likely this is, is to look at where his career mean performance is.

 

[Pirate55]

id say it's probably something taken in to huge consideration when assessing the risk v reward of any potential signing.

seems like common sense really : is this player on an upward curve or has he hit his peak?

Of course its common sense. No statistical theory of randomination required! There are no crystal balls, no guarantees, no absolute givens.

As I keep saying, football is not an exact science there are too many variables. Lots of opinions. Lots of views. Lots of excitement. Lots of upsets.

That is what keeps us all coming back!!!!

 

[Pirate55]

I think that you are either misunderstanding what you are reading or cherry picking the bits that you think supports what you are saying.

It is possible that Benteke could go on and become a great player. I have not seen anyone argue otherwise. What is being argued about is how likely is this? One way that we can get an idea of how likely this is, is to look at where his career mean performance is.

And what I am saying is that he doesn't have a long enough career at the age of 23/4 to be able to ascertain where his "career mean performance" is. Could we with Bale when he joined us?

 

[Bedfordspurs]

The problem with the regression logic in football is that it doesn't take into account experience, fitness and age.

If you want to apply it you can write off Costas only good season before this one. Yet he really does look the real deal doesn't he???

You cannot apply mathematical rules cart Blanche in a sport with so many variables including some of the above, plus who they are playing with and for (more chances/teams defending more/defensive teams), weather (yep... Some players play better in hotter/colder climates), and then you get the softer ones such as motivation and drive

I get the logic, hell I use it a lot when forecasting and making reccomendations to clients but I would not use it as a rule in sport

 

[Yiddo]

Have we even been linked to him?