Machine learning in trading: theory, models, practice and algo-trading - page 1629

 
Maxim Dmitrievsky:

SA?

It's all there about converting rows. So that's basic, what are you suffering from?

SA == StackOwerflow

There's nothing in there about fractality in the context of learning AMO.

It's horribly simple but it should work, it doesn't matter which rows you put there, just minutes/weeks, all together, it will find both patterns and the main thing is that it will give me adequate answer...

 
mytarmailS:

SA == StackOwerflow

There's nothing about fractality in the context of learning AMO

I invented a transformation, it's horribly simple but it should work, it doesn't matter what rows you put into it, minutes / weeks, all together, it will find both patterns and most importantly give an adequate response

don't worry about fractality, it's just a fiction.

No, I didn't mangle anything there
 
Maxim Dmitrievsky:

Forget about fractality, it's a fiction.

It's a fiction to look up the dimensionality by khurst, but scaling the data to one template for AMO is correct and necessary otherwise you just won't find repetitions in the data, and therefore no statistics, probability...

 
mytarmailS:

It's a fiction to look up the dimensionality by khirst, but scaling the data to one template for AMO is correct and necessary, otherwise you just won't find any repetitions in the data, and therefore no statistics, probability...

I scaled it, it's bullshit ) and by khirst bullshit, of course, and by entropy

 
Maxim Dmitrievsky:

I scaled it.

how?

 
mytarmailS:

how?

affine preconversions.

 
mytarmailS:

mikha! are you going to answer my question on the last page or not? did i get your article right?

I will answer it for sure. A little later. I went to change the refrigerator. Just for a couple of hours...
 
Maxim Dmitrievsky:

affine transformations.

let me guess, and you did all this in a sliding window, fixed size of course ? )

 
mytarmailS:

let me guess, and you did all this in a sliding window, fixed size of course ? )

the bigger the window, the better the correlation

and for different TFs did. It's all bullshit.
 
Maxim Dmitrievsky:

in the expanding one, like the bigger the window and the higher the correlation the better

Hmm, and the predictions on the training data, as well as on the recognition test data, did you also expand/decrease ? or there was a fixed mark