Machine learning in trading: theory, models, practice and algo-trading - page 2444

New comment
 
Renat Fatkhullin:

Complex complex numbers have been working for us for a couple of months now.

Thank you!
 
elibrarius:

Is it used anywhere in the MO? Is it an electrical engineer or something?

To be honest, I haven't met any problems, which can't be solved in pure MQL without any add-ons. But with complex numbers, yes, there was such a problem.
It's very nice to have them now.
Not electro, but micro and radio ;)
In order to make MQL finally work properly, the historical data should be divided into trend and flat segments. Complex numbers are very helpful in this process.
Personally, that's how I picture it.
 
Renat Akhtyamov:
That's my personal view.

How can they separate a flat from a trend?

 
mytarmailS:

how can they separate a flat from a trend?

If it works, I will show it to you. I haven't tried it yet.
Complex numbers are very useful when it comes to angle and direction, i.e. vectors and vector lengths, which in turn are these very segments that are currently lacking in almost any kind of theanalysis, including econometrics.
This is a very big topic to study, actually, a very unploughed field, so to speak.
 
mytarmailS:

How can they separate the flat from the trend?

During a flat, the trend is in the imaginary part and vice versa :-) you can not separate, they are together...

a joke of course...but as they say, only with a fraction

 

I decided to write about the existence of such a thing as FDA (Functional Data Analysis). This is a section of matstat and MO, where functions rather than numbers are used as features and/or outputs. Personally, I am interested in being able to get a probability distribution function as an output.

Also, more than once I've noticed people trying to "invent" this thing from scratch - usually in the context of solving the problem of finding some kind of "patterns" in prices).

If anyone is interested, there is a corresponding section in the CRAN Task View.

 
Aleksey Nikolayev:

I decided to write about the existence of such a thing as FDA (Functional Data Analysis). This is a section of matstat and MO, where functions rather than numbers are used as features and/or outputs. Personally, I'm interested in being able to get a probability distribution function.

Also, more than once I've noticed people trying to "invent" this thing from scratch - usually in the context of solving the problem of finding some kind of "patterns" of prices).

If anyone is interested, there is a corresponding section in the CRAN Task View.

It would be interesting, if you can explain it)))

 
Aleksey Nikolayev:

I decided to write about the existence of such a thing as FDA (Functional Data Analysis). This is a section of matstat and MO, where functions rather than numbers are used as features and/or outputs. Personally, I am interested in being able to get a probability distribution function as an output.

Also, more than once I've noticed people trying to "invent" this thing from scratch - usually in the context of solving the problem of finding some kind of "patterns" in prices).

If anyone is interested, there is a corresponding section in the CRAN Task View.

So what is the coolness of this method? Can you put it in simpler words?
The output in the form of a function or anything else you can get from a neural network, for example, what is the edge of this technology?
 
Aleksey Nikolayev:

I decided to write about the existence of such a thing as FDA (Functional Data Analysis). This is a section of matstat and MO, where functions rather than numbers are used as features and/or outputs. Personally, I am interested in being able to get a probability distribution function as an output.

Also, more than once I've noticed people trying to "invent" this thing from scratch - usually in the context of solving the problem of finding some kind of "patterns" of prices).

If anybody is interested, there is a corresponding section in the CRAN Task View.

covariation?
 
mytarmailS:
So what's the coolness of the method? Can you put it into simpler words?
The output in the form of a function or whatever can be obtained from a neural network, for example, what is the ejit of this technology?

More organic work with potentially infinite-dimensional objects.

1...243724382439244024412442244324442445244624472448244924502451...3492
New comment
Reason: