ForumSections

Machine learning in trading: theory, models, practice and algo-trading - page 2682

New comment
Serqey Nikitin #:

A limited perception of reality!

If we consider TRADING as a BUSINESS, then the issue of protecting the business and its secrets comes to the forefront....

And there is nothing to argue about: defence of your original methods of making a Profit - above all!

Don't confuse God's gift with eggs.
Sergey Gridnev #:
Don't confuse God's gift with an egg.
Don't get upset, "God's gift" doesn't apply to you!
[Deleted]
mytarmailS #:

Yes I did, and in more ways than one.

a simple tester can be written, but something serious with multiple orders, multi-tools, stops, take-outs, etc.... you can't write it properly, and the main thing is to make it work fast.

It will not be convenient and more complicated than you think in connection.

Write everything on your erka, then you will spend another year transferring the bot code to the terminal, since you have never written anything in it.
mytarmailS #:
Thanks for the detailed answer...
But do the people need one secure *.ech5? You need it for your services, but not for the people....

The people need convenience, less change, etc...
People are ready to sacrifice speed, security in favour of convenience.

If we are talking about the audience of consumers and sellers, then of course we need "one secure *.ex5".

If we talk about research for ourselves without a practical possibility to sell the works of our research, then we can screw them with duct tape and make notebooks. We have support for Python programs and notebooks right in the terminal and editor.

When it comes to training models, no one is willing to sacrifice speed. Appetite comes with opportunity and resource capacity has long been a limit. Therefore, there is a hard game of reducing the accuracy of calculations and then correcting the loss of accuracy.


We see the following applications of our ML solutions:

  1. External research with transfer (ONNX or rewriting) of models into MQL5
  2. Initial development of solutions in MQL5

Transfer to MQL5 will allow to use robots in the trading strategies tester and trade without intermediaries.
.
Renat Fatkhullin #:

Thank you

Maxim Dmitrievsky #:
It's not going to be convenient and more complicated than you think in connecting

Write everything on your erka, then you will spend another year transferring the bot code to the terminal, since you have never written anything in it.

There are backtesters in rk that are complex as I need, but slow.

simple, fast ones I can write myself...

Rk has a connection with MT5-4, I don't need to write code in mcl.

Maxim Dmitrievsky #:
It turns out that layers can already be built through matrices and vectors and calculate the gradient on them, it remains to add a couple of optimisers like Adam and you can write a neural network to try.

Ah, I haven't understood yet how the NS weights will be stored

You can already make your own analysis systems, including neural networks.

We have implemented the foundation and will make the next step - implement ML engine.

Please do not get personal and stay within the technical discussion.

[Deleted]
mytarmailS #:

Rk has backtesters as complex as I need, but slow.

simple, fast ones I can write myself...

Rk has a link to MT5-4, I don't have to write code in µl.

Also like that to be honest, a lot of surprises. It's better to transfer everything.
Renat Fatkhullin #:

Please do not get personal and stay within the technical discussion.

Can Nikitin(https://www.mql5.com/ru/users/vnik) be banned for a week for his constant rudeness towards his opponents?

1...267526762677267826792680268126822683268426852686268726882689...3748
New comment