Machine learning in trading: theory, models, practice and algo-trading - page 2762

 
JeeyCi #:

these are not signs - they are the dynamics of the development of a process over time - a dynamic series ...

and dependencies are studied as stationary series....

(but time can also be called a sign - exogenous, the time factor adds dynamics).

you didn't get neither meow nor hair on the input, but you got smoothing of the trajectory - neural network doesn't care what you approximate - just dynamics always shows the result with a lag - exactly because it needs the time factor as a window to collect a sample and estimate the rate of change of the dependent variable from time ... BUT the dependence (on time) must be there to analyse the dynamics (it is it that you put into the model you describe - if you call things by their names "what is the factor and what you want to know/evaluate" in the model - there will be less scribbles in (non)understanding each other on the forum)...

linear equation - shows velocity (tangent at a point to the curve of the trajectory), quadratic (parabola) will also show acceleration... and convergence of (f-a)^2 will be evaluated in time and will show the result on a finite segment of this time window - MLE (maximum likelihood estimation) always works the same way, at least when approximating statics, at least when equalising dynamics

unless you think about what you are looking at - a factor (qualitative/quantitative) or its dynamics (+ time factor) - you cannot distinguish dependencies and patterns of development - and therefore you do not understand what you are analysing and whether it is what you really need and what depends on what... and limitations of the type of analysis - analyses of dynamics ALWAYS show results with a lag

really, wearying arguments about who looks at what crookedly and sees what crookedly and interprets it crookedly, and how crookedly he himself understands from his interpretations and tries to convince others, and some in the posts above even with foaming at the mouth.... what kind of scientific dispute can we talk about? if you abstract everything and everyone to such an extent that you twist meanings with your freedom of speech -- there is no freedom of speech in natural sciences! there are exact formulations and their exact meanings ... not pseudoscientific knowledge ... not pseudoscientific knowledge, which you are promoting here because of your ignorance of basic fundamentals (and trying to present it as an argument)_

you create such models (curves) - not knowing what to put on the output (what you want to know) as a result of modelling ... and on what factors you are interested in this dependence

everything is too subjective often on this thread, so it is impossible to get to objectivity (which is the true and main goal of modelling).

Thank you for the detailed answer.

 
Has anyone tried the examples from the RL articles on this site? Q learning, actor-critic.
Does it work or not?
 
Ivan Butko #:

Are indirect signs possible? For example, cats and dogs often fight, but dogs are more likely to chase cats. We have two objects and their movements. The task: to determine which of them is a cat and which is a dog, having checked once by factual data and in subsequent times independently determine who is who. We know for sure that one of them is a cat and the other one is a dog, but we can't see their silhouette or hear them, we can't even see their traces, only the coordinate of movement. We feed the neural network the movement of objects back and forth (BUY-SELL). In the process of "thinking" and multiplication of weights, the neural network classified us that one object is always running ahead and the other behind it (MA_5[0] > MA_10[0]), and made an assumption: is the dog moving ahead now? Checked it with the actual data, got the answer (NO), corrected the data, assumed it was a cat, checked it - (YES). Now the neural network knows how to determine who is a cat and who is a dog by the fight and movement of objects. At the same time, it was not given paws, pieces of hair, teeth, barking or meowing.

That is, it seems that the neural network can be fed a lot of things and it will find something and find it in such a way(Hercule Poirot) that it will give the necessary answer. That is, the feature in this case does not represent partial information about the object being classified, but a solution is possible.

You can, but you should at least separate cats from dogs to begin with

About MAshki is a bad comparison, you need to clearly understand the difference between buy and sell marks. That's why it's called learning with a teacher. It won't do anything without a teacher. The neural network will just help you evaluate the correctness of your conclusions on new data.

You see how simple it is. Just look at the definition of what learning with a teacher is.
 
JeeyCi #:


really, wearying arguments about who looks at what crookedly and sees what crookedly and interprets it crookedly, and how crookedly he himself understands from his interpretations and tries to convince others, and some in the posts above even with foam at the mouth.... what kind of scientific dispute can we talk about? if you abstract everything and everyone to such an extent that you twist meanings with your freedom of speech -- there is no freedom of speech in natural sciences! there are exact formulations and their exact meanings ... not pseudoscientific knowledge ... not pseudoscientific knowledge, which you promote here because of your ignorance of basic fundamentals (and you try to present it as an argument)_

you create such models (curves) - not knowing what to put on the output (what you want to know) as a result of modelling ... and on what factors you are interested in this dependence

everything is too subjective often on this thread, so it is impossible to get to objectivity (which is the true and main goal of modelling).

Really tiresome arguments in different languages)))))) It would be good without emotions, with a literal and lucid explanation of understanding.... but as usual it is rare for people to understand that people are different and understand many similarities differently. ))))

Precise formulations and exact terminology are not possible at the front line of any science-like fields (at the stage of new research) unfortunately, that's why explanations of one's understandings in this thread are crucial for the results of here holivars)))))

 
Valeriy Yastremskiy #:

Really tiresome arguments in different languages)))))) It would be good without emotions, with verbatim and lucid explanation of understanding ... but as usual it is rare for people to realise that people are different and understand many similarities differently.))))

Precise formulations and exact terminology are not possible at the front line of any science-like fields (at the stage of new research) unfortunately, that's why explanations of one's understandings in this thread are crucial for the results of here holivars)))))

There is no advanced here, there is NOT knowledge of available tools. An advanced can be formed after mastering the available tools and trying to eliminate the shortcomings identified in them.


Absolutely precise formulations are possible in all cases when terms are supported by code.


For example, neural network is a generalising term with no specific content.

A neural network in the nnet package, on the other hand, has absolutely precise content.

Similarly, the term "predictive ability of the predictor", used by me and some other authors, is also a generalising term, but "predictive ability", understood as the difference between the medians between two vectors obtained by dividing the predictor by classes, is absolutely precise.

 
Maxim Dmitrievsky #:
Traits are what is fed to the NS input, and class labels are fed to the output.

A feature should represent partial information about the object being classified, that's what a feature is. An insignia, if you will.

The way I see it, as long as it is not defined what exactly is being classified, then all these 100 fancy ways of fitting will give the same result

The features are the real causes among the noise (not the features we need) that we are looking for, the class labels are the results we need, I agree with Sanych: The target-predictors pair are related and the pair exists precisely because they are related.

And they are hard to find)

Attributes are increments in time among other increments in time, or in a series, just in order, without taking time into account. The increments can be considered both separately, relative to the previous ones, and as a figure or function or pattern of consecutive increments. In order to find them, we examine a plot, a window, but for different features we need different sizes of this window. Sanych's approach is to do it on each new data result, yours is only on the one identified as necessary.

Also class labels / results we need, so I understand, to find the right ones is a separate task.

I think I understand everything correctly))))?

 
Valeriy Yastremskiy #:

The signs are the real causes among the noise (not the signs we need) that we are looking for, the class labels are the results we need, I agree with Sanych: The "target-predictors" pair are related and the pair exists precisely because they are related.

And they are hard to find)

Attributes are increments in time among other increments in time, or in a series, just in order, without taking time into account. The increments can be considered both separately, relative to the previous ones, and as a figure or function or pattern of consecutive increments. In order to find them, we explore a plot, a window, but for different features we need different sizes of this window. Sanych's approach is that it should be done on each new data result, yours is only on the one identified as necessary.

Also class labels / results we need, so I understand, to find the right ones is a separate task.

I think I understand everything correctly)))))?

We need to define classified objects with their attributes. What is a buy or sell trade and what are its attributes. How they differ.
 
СанСаныч Фоменко #:

There is no cutting edge here, there is NOT knowledge of available tools. A cutting edge can be formed after mastering the available tools and trying to eliminate the deficiencies identified in them.


Absolutely precise wording is possible in all cases when the terms are supported by code.


For example, neural network is a generalising term without specific content.

But neural network in the nnet package has absolutely precise content.

Similarly, the term "predictive ability of the predictor", used by me and some other authors, is also a generalising term, but "predictive ability", understood as the difference between the medians between two vectors obtained by dividing the predictor by classes, is absolutely precise.

I'm not going to argue, I agree that everything in the code is absolutely accurate.

Of course, not knowing or having different levels of knowledge of the tools does not allow full understanding between participants. But also categorical requirements to study them are not always good for full understanding. Besides, the subject of the dispute/holivar can be defined precisely enough.

 
Maxim Dmitrievsky #:
It is necessary to define classified objects with their attributes. What is a buy or sell deal and what are its attributes. How they differ.

The signs of deals are their results, properties of deals are still there, but the signs for making a decision on deals are increments and time (I do not like the serial number).

 
Valeriy Yastremskiy #:

The signs of transactions are their result, the properties of transactions are still there, but the signs for making a decision on transactions are increments and time (I don't like the ordinal number).

The signs of the object: tail, whiskers, ears. Object cat. Generalise to all objects of the same type. Signs to sell: ... ... ... . Generalise. If they are not generalised, then either the signs are wrong or the object under study.