Machine learning in trading: theory, models, practice and algo-trading - page 1903

[Mihail Marchukajtes]

Maxim Dmitrievsky:

I guess... I just don't understand what you want from me ))

I only code my own crazy ideas, I have no time for the rest

I'm not asking you to code I just wanted to send me a pre-processed file of the data, that you would have built the model and I would have run it at home. Aren't you interested in the results? Maybe you'll get something interesting?

[Valeriy Yastremskiy]

Aleksey Nikolayev:

Are there any significant differences from the one used in your Expert Advisor?

https://www.mql5.com/ru/forum/241140/page6#comment_16341424

I have not tested it myself. But it has surprised me.

[Mihail Marchukajtes]

You can do it the other way around. I have a trading Expert Advisor which takes the result of the model from the library. The optimizer forms a library with a trained model. You can send me your training file with only data and no information about inputs. I train your file and send you the model. You send it to MT5 and look how the model works on data that it has not seen. How about this?

In other words, the secret of getting the input data is on your side and the model that I received cannot be used because I do not have any inputs. As a result, you write the input transformations in the library and get a working model.

HOOOOOOOOO? Anyone interested in????

I'll even run a shaitan machine for it.....

[Maxim Dmitrievsky]

Mihail Marchukajtes:
I'm not asking you to code, I just wanted to send me a pre-processed file of my data so you could build a model and I could run it on my own. Aren't you interested in the results? What if something interesting comes up?

I write thousands of lines of code... No, I'm not interested in results, I'm interested in valid tried-and-true ideas

without any water.

I haven't heard any in the last few years or so.

I don't make a mystery out of anything because it's stupid absurd.

[Aleksey Nikolayev]

Valeriy Yastremskiy:
https://www.mql5.com/ru/forum/241140/page6#comment_16341424

I haven't tested it myself. But it surprised me.

Yes, but with your version is also a problem.

[Mihail Marchukajtes]

Maxim Dmitrievsky:

I write thousands of lines of code... No, I'm not interested in results, I'm interested in valid tried-and-true ideas

without any water.

I haven't heard any in the last few years or so.

I don't make a mystery out of anything, because it's a stupid absurdity

Well, send me the training file with dirty data and I'll try to clean it up and get a model. Can you do that?

Then upload the model to yourself and test it on a different interval.... Within 500, maybe 1000

[Maxim Dmitrievsky]

Mihail Marchukajtes:

Well, then send me the training file with dirty data, and I'll try to clean it and get a model. Can you do that?

Then you can upload the model to yourself and test it on another interval.... I'd say 500, maybe 1,000.

Here, have some fun.

errors on a simple tree.

1.0 0.8648648648648649

you can do better, well done.

[Valeriy Yastremskiy]

Aleksey Nikolayev:

Yeah, but there' s a problem with your version, too.

If you do not know anything about this, look at the chart. They are different. )

The essence is not about that. Normal archive of news, with an assessment of the strength of the news found only on the eurodollar. I could not find others. I do not know how to take it from there.)

[Aleksey Nikolayev]

Valeriy Yastremskiy:
Yes, it parses not with investings, but with efixfactory. They are different. )

It seems to be the same forexprostools that you have, taken from investingsom widget.

[Aleksey Nikolayev]

Valeriy Yastremskiy:

The essence is not about that. Normal archive of news, with an assessment of the strength of the news found only on the eurodollar. I could not find others. There is also a variant for different currencies, I do not know how to get from there without a rake)

I do not know what to expect.)