"New Neural" is an Open Source neural network engine project for the MetaTrader 5 platform. - page 95

 
Dmitry Fedoseev:

Then the real question is why aren't people using it? Is there something wrong with it?

Some people are using, they write in their private messages

but no one knows how to pick out patterns in the market, i.e. on, almost, SB. And this is a major hitch for any ML strategy, and it's something to think about rather than anything else.

 
Реter Konow:
Asking experts. What is a "neuron" in the programmatic sense? How much does it resemble an ordinary function? There is also input, transformation and output of values. What is the difference between a "neuron" and a function?

The function is. It rises sharply at first, then slowly. Sort of like a logarithm. And the synapse is the input. Axon is the output. (or vice versa))

 
Maxim Dmitrievsky:

Some people use it, they write in person

but no one knows how to isolate patterns in the market, i.e. in, almost, SB. And this is the main bottleneck for any ML strategy, and it is something to think about and nothing else.

If everything rests on SB, then nothing will work.

 
Dmitry Fedoseev:

If it's all about SB, then nothing will work.

nothing works for anyone for that reason

 
Dmitry Fedoseev:

The function is. It rises sharply at first, then slowly. Sort of like a logarithm. And the synapse is the input. Axon is the output. (or vice versa))

I wasn't referring to a mathematical function, but a software function. A neuron, as a software entity, is conceptually different from a software function entity that takes parameters, converts them and outputs the result.
 
Dmitry Fedoseev:
What's there to talk about? When a genetic algorithm contest was scheduled, 90% of the forum got hysterical, like that's possible, it turned out that nobody knew what it was, not to mention having their own implementation, or at least trying to make one... but hey, how much fun everyone had!

yeah... it was an epic fiasco... in every way... I'm probably a lousy organizer, too.

 
Maxim Dmitrievsky:

Some people use it, they write in person

but no one knows how to isolate patterns in the market, i.e. in, almost, SB. And this is a major bottleneck for any ML strategy, and it's something to think about and nothing else.

Maxim, until the time between BP values is fed to the input of the neural network and the grid calculates the market cycles (which it has, I assure you) - nothing will work. You should search on samples from one trading session up to one year. The sample shouldstrictly correspond to the time period and nothing else.

It is in the time structure that market BP differs from SB, I have written this many times before.

 
Maxim Dmitrievsky:

nothing works for anyone, for that reason

Let's say "Almost doesn't work"... and that's what's left of the whole "Doesn't work" and is left to scrape together the scraps.

 
Реter Konow:
I was not referring to a mathematical function, but a software function. A neuron, as a software entity, is conceptually different from a software function entity that takes parameters, converts them and outputs the result.

It's no different. It is an ordinary function. Input one parameter, output one value.

Искусственный нейрон — Википедия
Искусственный нейрон — Википедия
  • ru.wikipedia.org
Схема искусственного нейрона 1.Нейроны, выходные сигналы которых поступают на вход данному 2.Сумматор входных сигналов 3.Вычислитель передаточной функции 4.Нейроны, на входы которых подаётся выходной сигнал данного 5. — веса входных сигналов Иску́сственный нейро́н (математический нейрон Маккаллока — Питтса, формальный нейрон[1]) — узел...
 
Alexander_K:

Maxim, until the time between BP values and the grid calculates market cycles (which it does, I assure you), nothing will work. You should search on samples from one trading session up to one year. The sample shouldstrictly correspond to the time period and nothing else.

It is in the time structure that market BP differs from SB, I have written that many times.

Exactly so, but I have not yet formulated a general approach from an MoD perspective to this :)

By the way, could it work as a substitute for Hearst?https://en.wikipedia.org/wiki/Sample_entropy

Or also a lagging one.

Sample entropy - Wikipedia
  • en.wikipedia.org
Sample entropy (SampEn) is a modification of approximate entropy (ApEn), used for assessing the complexity of physiological time-series signals, diagnosing diseased states.[1] SampEn has two advantages over ApEn: data length independence and a relatively trouble-free implementation. Also, there is a small computational difference: In ApEn, the...