Machine learning in trading: theory, models, practice and algo-trading - page 1910

[Mihail Marchukajtes]

Rorschach:

Thank you, interesting, it would definitely not occur to me to feed the previous signals. This begs the question, maybe the network is an unnecessary link?

That is why the network is interesting to me, so as not to select the signs. Otherwise it would be easier to make a classic system.

I will have to tinker with some more experiments, I will take a very large number of examples of the system on the wrist, so that when training there is not a single repetition, and it was possible to fit into one epoch.

Unfortunately the wizards won't help you alas. Where are you going? What is this place called?

[Mihail Marchukajtes]

Maxim Dmitrievsky:

Adabust gave me 79 for 79

So, Maxim, are you sulking? You got warmed up by the quality model and now you're chipping away at the right and left???? That's the thing, we are all IMPORTANT when we are sitting on the white throne. I now plan to form a team and now I've decided that no one here is a tete-a-tete. Wrong guy attacked.... :-)

[Maxim Dmitrievsky]

Mihail Marchukajtes:
So, Maxim, are you sulking? You got warmed up by the quality model and now you're chipping away left and right???? That's the thing, we are all IMPORTANT when we are sitting on the white throne. I now plan to form a team and now I've decided that no one here is a tete-a-tete. Wrong guy attacked.... :-)

It's not working, it's just more bullshit.

but it was fun.

[Mihail Marchukajtes]

Maxim Dmitrievsky:

It doesn't work, it's just more bullshit.

but it was fun

Do you know why? Let me now go record a video and there will explain everything. There's a preface, and you're welcome to trade the PLO of your choice and post the result. Okay?

[Rorschach]

Mihail Marchukajtes:
Unfortunately, unfortunately, the mashkas will not help you. Where did you come to? What is this place called?

Run-in wrenches.

More than once I met the opinion that if there is a pattern, it is easily pulled out by any methods.

[Mihail Marchukajtes]

Rorschach:

Run-in mash-ups.

More than once I've encountered the opinion that if there is a pattern, it's easy to get it out by any method.

I agree, but it doesn't exist in mashcats... And don't argue.... Pseudo pattern yes, but that's at the level of chance, and in the market it's the stability that counts.

[Mihail Marchukajtes]

As promised. In some places with pauses and terrible sound. I recommend watching with headphones. With a personalized message to Maxim. Maybe you will find the answer in this video, if you accept the criticism constructively.

Watch and do not hesitate:-)

[mytarmailS]

Mihail Marchukajtes:

As promised. In some places with pauses and terrible sound. I recommend watching with headphones. With a personalized message to Maxim. Maybe you will find the answer in this video, if you accept the criticism constructively.

Watch and do not hesitate:-)

))))

I like it.)

[Maxim Dmitrievsky]

Mihail Marchukajtes:

As promised. In some places with pauses and terrible sound. I recommend watching with headphones. With a personalized message to Maxim. Maybe you will find the answer in this video, if you accept the criticism constructively.

Watch and do not hesitate:-)

You're making it up as you go along.) 24 columns, not 100. You asked for the file yourself. There are no mistakes (I explained). 300 and a lot of lines because I gave you a year, so that your 'generator' did not pooped count))) But go on. I didn't have time to finish watching it, but the beginning is promising. I'll leave a full review later. Looks like I'll have to respond in video format.

[mytarmailS]

Maxim Dmitrievsky:
But keep going

Yeah, watching you is much more interesting than reading...

By the way you can compress the number of signs even before the Vtree (or after) by the same PCA with almost no loss of quality.

here's an example how from 1000 variables you can make 10 , with no information loss at all and it's the full equivalent of those 1000 signs

#  data set Х
X <- matrix(rnorm(10000) , ncol = 1000,nrow = 1000)
colnames(X) <- paste0("V" , 1:ncol(X))
head(X)

library(caret)
pcm <- preProcess( X[1:500,] , method = "pca" ,thresh = 1)
PCs <- predict(pcm , X)
head(PCs)

thresh = 1

means you want to save 100% of the information

thresh = 0,7 

70%, etc.

=================================================

In the video you said you want to sample only unique values but you don't know how -

just cluster the values in the sample and remove repetitive clusters

You can cluster by rows, columns or even every value