Artificial neural networks. - page 2

New comment
 
It would be great if neural network carifes such as Urain, TheXpert, MetaDriver, sergeev could come on board.
 
joo:
It would be really great if such neural network carphees would pull up here...

Have any of them (not the personalities, but the 'coryphaei') made a fortune, or only in theory???

 
St.Vitaliy:

Have any of them (not the personalities, but the "coryphaei") made a fortune, or only in theory???

At least there are challengers who have won the championship among those who have used neurocells.

Or do you want to present something to me or to my good friends or acquaintances?

 
joo:

At least there are challengers who have won championships among those who have used neural nets.

Or do you want to present something to me or to my good friends or coryphaei?

A person can only teach or advise by the place, something he or she has spent a lot of time doing.

A good programmer is brilliant at writing programs, if that is your ultimate goal, then it is all very logical. But it has very little to do with the market.

 
St.Vitaliy:

A person can only teach or advise you about something they spend a lot of time doing.

A good programmer is brilliant at writing programs, if that is your ultimate goal, then it makes perfect sense. But it has very little to do with the market.

Thank you. Now, please, translate into Russian. I do not speak your dialect, sorry, I do not sprechenois.

I just do not understand whether I am a bad (or vice versa) programmer, or I do not apply neural networks to the market, or something else. Please explain.

 
joo:

Thank you. Now, please translate into Russian. I don't speak your dialect, sorry, I don't sprech.

I just don't understand whether I am a bad (or vice versa) programmer, or whether I don't apply nerves to the market, or something else. Explain, please.

Believe me, I don't use value judgments on people. At least I try very hard to do so, only facts.

What is the basis for thinking that neural networks really do allow you to earn steadily, the corifes have a successful practical experience that cannot be attributed to a price artifact?

It just all looks very pathetic, but is there a rational grain to it?

But the simple idea (as an example) of buying sugar in autumn and selling in early summer is not interesting...

 
07041982:

Neural networks are IMHO the hardest part of algorithmic trading. There are no newbies here. I, too, only after 8 years of forex, having spent a lot of money, having tried all possible strategies and indicators and all possible combinations of them, have decided that this is the only trading strategy I have not touched. Why have I not tried it before? I thought it was too complicated and required special programs like "neurosolutions". The most important thing is to understand how neural networks work, and the rest is a technical matter.

I invented my first prototype neuronet in about a day, wrote it in 15 minutes using pure MQL5, trained it using optimization. And oh my miracle, the results are encouraging... tank robustness is present for several years, tank tests are stable. Now I'm almost sure that if the Grail exists, it is hidden somewhere in the depths of neural networks: their potential is almost unlimited, I can add there any number of indicators with any number of parameters... and it can be done using pure MQL5. After thinking and trying them, all my previous efforts to create an Expert Advisor out of standard indicators appear to be child's play.

In this thread I propose to share my experience in the field of neural networks. It would be interesting to learn the principles of designing and training networks in MQL5. I am waiting for your comments.


As far as I can judge no one here has implemented neural network with non-permanent architecture. A little explanation, one neuron can connect with 5 - ..... other neurons lying in the next and previous layers. Thus, the notion of a layer is smeared out a bit like a neural network with feedbacks, but these links are constantly changing along with weight coefficients, besides the number of neurons in the network is not constant they die off and recover. It is such an analogue of a natural neural network. Judging by publications in the network, it is a more promising direction.
 

The advantage of networks over decision trees in forex trading is not obvious. It is foolish to use networks wherever you want. Apply them wherever a non-linear output function from an input may exist. Let me give a simple example of classification. Suppose we need to create a system that distinguishes different mammals. The input is information about the shape and size of the whole body and its parts (head, nose, ears, paws, tail, etc.) Do you really think that creating a non-linear model like Class = F (mustache, paws, tail) will be more accurate in determining the class of the mammal than a decision tree mustache? paws? tail?

 
NSs can find hidden dependencies such as patterns, levels, etc. plus they can cram a lot of input parameters in, using conventional methods you don't have the power to optimise so many parameters
 

07041982:
НС могут сами находить скрытые зависимости типа паттернов, уровней и т.д. плюс в них можно запихнуть очень много входных параметров, с использованием обычных методов вам не хватит мощностей столько параметров оптимизировать

The network is a tool, a kind of universal non-linear function that can be optimised (fitted) for different data (input-output). This function cannot pick up any regularities. You may as well argue that a polynomial fitted to a smooth curve finds its hidden patterns. Step out of the data range to which the function has been fitted and you know what you will get as an example of polynomial regression. By the way, the number of optimisable parameters in the network is much larger than in the decision tree as the network contains sub-connections which do not influence to correct decisions and whose weights are decreased during optimisation.

123456789...14
New comment