Machine learning in trading: theory, models, practice and algo-trading - page 382

[Maxim Dmitrievsky]

Vasily Perepelkin:
A hahahahaha! Timofey Martynov is a leading trader in Russia, like Gerchik or Anatoly Radchenko. Talking about Timofey's hand-wringing is like talking about Soros' hand-wringing, which in turn shows your qualifications as a trader.

That's it, go to kindergarten.

[Ivan Negreshniy]

SanSanych Fomenko:

What machine learning algorithm should be used?

Useful material, a simple and clear cheat sheet, and most importantly the right one and when used, at the stage of problem definition, can save a lot of time and money.

For example the fact that random forests in classification tasks, and especially regression, do neural networks in terms of accuracy and speed, I personally convinced, only on experiments.

[Maxim Dmitrievsky]

Ivan Negreshniy:

Useful material, simple and clear cheat sheet, and most importantly correct and when used, at the stage of problem definition, can save a lot of time and money.

For example the fact that random forests in classification tasks, especially regression, make neural networks in terms of accuracy and speed, I personally made sure only on experiments.

https://msdn.microsoft.com/ru-ru/mt745082.aspx more here

[Ivan Negreshniy]

Maxim Dmitrievsky:

https://msdn.microsoft.com/ru-ru/mt745082.aspx more here

Thanks - Microsoft, of course, is broader and half a year ahead of SAS, but in essence the results are similar.

[forexman77]

How are "forests" different from neural networks?

[mytarmailS]

Honestly speaking, I read this branch every once in a while, and what they write here is mostly a sift or reference to some articles, no own experiments and codes, the branch has rotted((( and does not carry any useful load on the brain, here maximum 2-3 people understand what they write, unfortunately... And the rest of the crowd is stupidly "copy-pasting" articles, codes and even someone else's thoughts, not suspecting that this is a scam, and there is no possibility, desire or ability to check it ...

[Maxim Dmitrievsky]

mytarmailS:
Frankly speaking, I read this thread every once in a while, and what is written here is the vast majority of siftings or references to some articles, no own experiments and codes, the thread has rotted((( and does not bear any useful load on the brain, here most 2-3 people understand what they write, unfortunately ... And the rest of the crowd is stupidly "copy-pastes" articles, codes and even someone else's thoughts, not suspecting that this is a fake, and there is no opportunity or desire or ability to check them ...

you do not even understand what is written in these articles, why do you need the codes) it was a rhetorical question that does not require an answer

[Alexander Ivanov]

The branch is normal)

The main thing is not to get bored)).

[Ivan Negreshniy]

forexman77:
What is the difference between a "forest" and a neural network?

As a programmer I can explain to you the differences in the source code. While neural networks in source code usually look like data, i.e. arrays of weight coefficients, plus iterative algorithms for their processing, decision trees, which make up a forest, are converted into pure code, such as functions in MQL with variable comparison operations, conditional operators, which are called hierarchical, branching algorithms...

[Yuriy Asaulenko]

Maxim Dmitrievsky:

So you don't even understand what's written in the articles, why do you need the codes?

I do not even understand what the codes are for.) The essence of anything is faster and easier to set forth in ordinary words. And the codes we ourselves are able to write.)