Machine learning in trading: theory, models, practice and algo-trading - page 958

 
Alexander_K2:

One way or another, but the branch has reached another dead end, thanks to the efforts of the ubiquitous Asaulenko and SanSanych (like mine, but with their own characters).

I could yell in falsetto: "Alyosha, Koldun, well, have pity on us wretched! Give us a drink from the Grail! My soul is burning."

The magician has already admitted that he didn't drink from the grail, but only took a sip.

There are strategies, but they are not kept secret, it's just not customary to talk about them so as not to attract too much publicity and not to spoil the nature of liquidity in the market, when the masses come to trade the same thing. I have already mentioned above nickname, having studied his posts and brief references to the MO, one can understand what he is talking about

The other characters, of course, are misleading and you should not listen to them seriously
 
Maxim Dmitrievsky:

The magician has already confessed that he did not drink from the grail, but only took a sip.

I have already mentioned above a nickname, whose posts with short references to MOs you can understand what he has done and what he has not drunk from the grail. I have already mentioned above a nickname, having studied his posts and brief references to the MO, one can understand what he is talking about

Yep. I get it. It seems to be the notorious hrenfx, he gets banned all the time :))

 

What do you think of the idea to use two models for mutual instruments, and if two models show a signal in the right direction, then enter the market?

I think the same approach can be used in training for different TFs of the same instrument.

 
Aleksey Vyazmikin:

What do you think of the idea to use two models for mutual instruments, and if two models show a signal in the right direction, then enter the market?

I think the same approach can be used in training for different TFs of the same instrument.

It's a good idea to introduce the notion of mutual instruments

How does this differ from the ensemble?

 
Maxim Dmitrievsky:

Aleshenka said that there are a lot of books on working with tics, but did not send any in my personal message

I understand - he just makes bars in 1 second and uses them to teach NS. The question is whether there are fishes in there, and whether there are no fishes in there.
The question is - is there a fish there?
 
elibrarius:
I guess he just makes bars at 1 second and uses them to study NS. Tick books are completely unnecessary - the local reference book is enough to convert ticks into 1 sec. bars.
The question is - is there a fish there?

I would add tick TFs to mt5, it could be useful for some people.

But 600 features is very strange for hft, i don't believe it.

 
Maxim Dmitrievsky:

to begin with, it would be a good idea to introduce the concept of reciprocal instruments

How does it differ from an ensemble? Right, nothing.

By reciprocal instruments, I mean instruments that correlate with each other, or have a non-linear dependence that cannot be detected by ordinary correlation.

Ensembles - I thought that this is one and the same sample, but different combinations of predictors.

 
Aleksey Vyazmikin:

By reciprocal instruments, I mean instruments that correlate with each other, or have a non-linear dependence that cannot be detected by ordinary correlation.

Ensembles - I thought it was the same sample, but different combinations of predictors.

Well not an ensemble but a committee then, on different samples, or a committee of ensembles

If the dependence is nonlinear then it's probably not there at all, or it's random. It is necessary to look for a linear

 
elibrarius:
I guess he just makes bars in 1 second and uses them to teach NS. You don't need any books on ticks at all - the local reference book is enough to convert ticks into 1 sec. bars.
The question is - is there a fish there?

There are fish there, but very small, but a lot of them - scalping is called. On the stock exchange it works well. But in forex, when the minimum spread is 1 (10) p, and it is not so fast, it's all mind games. In forex it's to save some money on entry and exit, and then you watch how this structure bounces back and forth.

 
Maxim Dmitrievsky:

I would add tick TFs to mt5, it could be useful for some people.

But 600 features is very strange for hft, i don't believe it

Maybe 600 second bars is the last 10 minutes...
Reason: