Machine learning in trading: theory, models, practice and algo-trading - page 44

[mytarmailS]

Alexey Burnakov:
A You can try it. Sometimes they remove the tails of distributions and sometimes it helps.

And sometimes they cut off their paws, too ;)

What are you talking about?

[Alexey Burnakov]

mytarmailS:

And sometimes they cut off their paws, too ;)

What are you talking about?

Learn! Paws, horns, hooves.

You can tell a lot about the process by the tails of the distribution.

http://www.long-short.ru/post/raspredelenie-s-tolstymi-hvostami-491

[Alexey Burnakov]

By the way, is anyone interested in this or not, I don't get it. Do you need a trained robot that passes validation in 5 years with a profit?

Like

I'm back from vacation. I can prepare the files and post them, and whoever needs it, will improve it for themselves.

[Andrey Dik]

Alexey Burnakov:

By the way, is anyone interested in this or not, I don't get it. Do you need a trained robot that passes validation in 5 years with a profit?

Like

I'm back from vacation. I can prepare the files and post it, and whoever needs it, will improve it for themselves.

Very Low Profitability.... It is more reliable to keep money in the bank.

[Alexey Burnakov]

Andrey Dik:

Very low profitability.... It's safer to keep money in the bank.

And I won't put it out with a high one.

I told you to work it out.

The main thing here is that it is profitable, not fit, as in the market. If not, then no.

[Andrey Dik]

Alexey Burnakov:
And I just do not put them on high.

It's understandable... )

I saw the status - a purely mechanical reaction... Somebody needs it.

[mytarmailS]

Alexey Burnakov:

Learn! paws, horns, hooves.

You can tell a lot about the process by the tails of the distribution.

Thanks, I read it, but it's not what I think it's about. Let me try to explain in a less abstract way...

We have a predictor, let's call it an rsi indicator, it has a value range from 0 to 1

1) divide it into 10 ranges

2) make something like predictors from these ranges

3) build a model on them, let it be random f.

4) look at the predictors and see that one of the 10 ranges is ten times stronger than the others

5) read the question I asked in the previous post ;)

I think there's no need in tails and distributions, right?

[Alexey Burnakov]

Andrey Dik:

It's understandable... )

I saw the status - a purely mechanical reaction... Somebody has to.

And purely mechanically it was wrong. The yield of 20% per year in dollars you just can't find. This is a calculation based on the FS.

[Alexey Burnakov]

mytarmailS:

Thanks, I read it, but I think it's not about that at all, I'll try to explain less abstractly...

We have a predictor, let it be an rsi indicator, it has a range of values from 0 to 1

1) divide it into 10 ranges

2) make something like predictors out of these ranges

3) build a model on them, let it be random f.

4) look at the predictors and see that one of the 10 ranges is ten times stronger than the others

5) read the question I asked in the previous post ;)

I don't need tails and distributions here, right?

How do you make multiple predictors from the ranges of one predictor? I don't understand it.

[Andrey Dik]

Alexey Burnakov:
And purely mechanically it is wrong. You won't find 20% return per year in dollars for nothing. This is a calculation based on FS.

Okay, that's a very good trading performance on history! Congratulations.