Discussion of article "The Box-Cox Transformation"

New comment
 

New article The Box-Cox Transformation is published:

The article is intended to get its readers acquainted with the Box-Cox transformation. The issues concerning its usage are addressed and some examples are given allowing to evaluate the transformation efficiency with random sequences and real quotes.

Fig. 1. The Box-Cox transformation in case of various values of the lambda parameter


Author: Victor

 
Victor, do you think that in the case of a poor approximation to normality after a BC transformation, is it advisable to reapply the same transformation?
 
denkir:

Victor, do you think that in case of poor approximation to normality after BC transformation it is reasonable to reapply the same transformation?

I don't know, but I think that reapplication of the transformation will no longer have as strong an effect as the first one.

It seems to me that this kind of transformations are not perfect. Application of such transformation, as well as any other, leads to change of initial characteristics of the input sequence (probably). And here the main thing is not to overdo it, otherwise the sequence obtained after transformations will have nothing in common with the original one. That's probably why transformations that can bring any input sequence to a normal one are not widespread. But let me emphasise once again, I have not seriously considered these questions.

 

I see. Yes, it's quite a deep subject. One can, as they say, saw and saw.....

The article is very informative. There is a logical connection with what you wrote earlier. Thank you for the material.

 
If we talk about trading, the stability of the characteristics of the quotient when moving along it is of interest. You have given the characteristics of change after transformation without shifting, but what will happen to the BC parameter when shifting one bar forward? If we compare the stat characteristics by shifting sequentially along the untransformed cotir with the stat characteristics of the transformed cotir, what do we see? Does the variance fluctuation decrease with the shift. If it does, that is a huge advantage for BC.
 
denkir:
I see. Yes, it's quite a deep subject. One can, as they say, saw and saw.....

The article is very informative. There is a logical connection with what you wrote earlier. Thank you for the material.

Thank you for the evaluation of my work.

 
faa1947:
If we talk about trading, the stability of the characteristics of the quotient when moving along it is of interest. You have given the characteristics of change after transformation without shifting, but what will happen to the BC parameter when shifting one bar forward? If we compare the stat characteristics by shifting sequentially along the untransformed cotir with the stat characteristics of the transformed cotir, what do we see? Does the variance fluctuation decrease with the shift. If it decreases, then that is exactly what is a huge plus for BC.

This article was intended as an entry level article, designed primarily to alert the reader to the features of classical statistical methods and to provide some sort of toolbox for experimentation. Your questions go far beyond the scope of this article. I will not be able to answer them for you.

New comment