Machine learning in trading: theory, models, practice and algo-trading - page 3657

New comment
 
Who cares, give me an example of learning without a teacher

What's going on? What is feeding such a network? What does it do to the dataset (prices)?
 
mytarmailS #:
First of all, you need to come up with a definition of flat/trend for yourself
And then it's a matter of technique
In the first case when trades against the trend give more profit, in the second case when on the trend. You can also use the Hurst index or average volatility.

I need it for label autopartitioner and for quick determination of what strategy is better to use.
 
Ivan Butko #:
Who cares, give me an example of learning without a teacher

What's going on? What is feeding such a network? What does it do with the dataset (prices)?

The simplest example: the input is some data (anything, several stochastic indicators, for example), the output of the network is trading signals. The target function is any custom criterion of trading results on history.

This method can neither be fattened nor started. And so, there are a lot of different ways to learn without a teacher, mostly concerning the topic of clustering.

 
Ivan Butko #:
Who cares, give me an example of learning without a teacher

What's going on? What is feeding such a network? What does it do with the dataset (prices)?
Google unsupervised learning. Clustering, compression. Only attributes as input, no labels.
 
Andrey Dik #:
The simplest example: the input is some data (anything, several stochastic indicators, for example), the output of the network is trading signals. The target function is any custom criterion of trading results on the history.
Sit down, stake
 
Andrey Dik #:
The simplest example: the input is some data (anything, several stochastic indicators, for example), the output of the network is trading signals. The target function is any user criterion of trading results on the history.

That is, the NS will arrange entry points by itself? Or classify patterns by itself?

I don't understand how to create a TS with the help of the NS "without a teacher".

How it will look like (will there be entries? How many? Will the NS decide what to enter?)

It's not clear at all.

 
Maxim Dmitrievsky #:
Google unsupervised learning. Clustering, compression. Only features as input, no labels.

I don't understand what the result is

Googled, the picture is of an aubergine that has been pulled out of the pile and put in the basket.

What about forex? Does it (it) even apply here?

 
Ivan Butko #:

That is, the NS will set the entry points by itself? Or classify patterns by itself?

I don't understand how to create a TS with the help of NS "without a teacher".

How it will look like (will there be inputs? How many? Will NS decide what to input?)

It's not clear at all.

Ivan, write in private. It is not comfortable to communicate here, among "teachers".
 
Ivan Butko #:

I don't get it

Googled, picture of an aubergine that was pulled out of the pile and put in the basket.

What about forex? Does it (it) even apply here?

Split the data into several similar groups, for example. Applicable, why not. Through clustering you can look for patterns. Each group will contain similar observations.
 
Andrey Dik #:
Ivan, write to me in private. It is not comfortable to communicate here among "teachers".

Andrey, if it is not difficult, describe how you imagine the implementation of "without a teacher" at Forex, in my topic"What to feed to the neural network input"

If you have a earning machine without a teacher - then in private

1...365036513652365336543655365636573658365936603661366236633664...3674
New comment