Discussion of article "Metamodels in machine learning and trading: Original timing of trading orders" - page 7
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Have you even read anything besides a book on mql5?
You don't realise how ridiculous you are just, but you can go on ) barging into a thread with your schizo statutes that have nothing to do with anything. Can't program, can't MO, can't do anything but pseudo-philosophy. Why do you need to be here? Did you read the thread title? I have it in my favourites, you're distracting me. I wonder how many times you'll get it?
I think that before you teach a machine, it would be good to learn to understand the process yourself.
What makes you think Plato's philosophy is "pseudo-philosophy"?
Actually, I'm trying to find at least one normal programmer who understands what he's doing.
That's what I think, before you teach a machine, it would be a good idea to learn to understand the process yourself.
What makes you think that Plato's philosophy is "pseudo-philosophy"?
Actually, I'm trying to find at least one normal programmer who understands what he is doing.
Go to freelance and out of this thread, you're tiresome
Good luck to you, too.
Hi Maxim, I really like your idea of translating models from Python to MQL5. Before that I used a dll that I wrote myself, it was very inconvenient, but through it I have tried Python, R language, ML.Net, Keras, H2O.ai, and many other things. I respect your experience in machine learning, but I think you lack the experience of a programmer in implementing your ideas. I'm a programmer myself, here's my https://www.linkedin.com/in/eugene-chistyukhin-245548127. I have improved your code on python also significantly improved EAs, wrote my own site for testing these robots https://forexvisible.com. Meta models, also threw on the tests, there and my and your developments with improved predictors and already as more than a year of tests, but they are not all got there, look if interested, write, I will be glad to answer. In two will be easier to take this rock, we will show them scam, write children ) do not pay attention
Hi, glad that you find something interesting for yourself. I seem to have realised all the ideas somehow :) Each article emphasises on one thought or idea, so you can borrow and improve them.
I haven't had time to write about signs yet, although there are interesting things, but I don't have a general concept of how to combine them into an article yet. Probably won't write it since these "kids" are trying to ban me :)
I don't have access to linkdin (long ago banned in the country). You can write here, in private or cart. And in general on the forum there is a big thread on MO.
Hi, glad you find something interesting for yourself. I seem to have realised all the ideas somehow :) Each article emphasises a single thought or idea, so it's just that they can be borrowed and improved.
I haven't had time to write about signs yet, although there are interesting things, but I don't have a general concept of how to combine them into an article yet. Probably I won't write any more, as these "children" are trying to zoban me :)
I don't have access to linkdin (long ago banned in the country). You can write here, in private or cart. And in general on the forum there is a big thread on MO.
Hi, yes about signs it would be interesting to read the article, if you find time and desire, very much waiting.
I tried different indicators as signs, I stopped on Bollinger Bands, but strangely enough with settings that make this indicator the same MA indicator that you use. I wanted to roll back to your code, but strangely enough Bollinger Bands showed better results, perhaps because I pass to the training data on 3 lines of the indicator, although they are repeated.
I use 28 trading pairs, and the most important thing is that every week I retrain the models on new historical data, I run all 28 pairs with new models in the strategy tester, and if a model has shown better results on one of the 28 pairs, I replace it in the Expert Advisor with a better one. I have automated this process, so it doesn't take much time. As a result, every week, especially in the first two months, the models are actively improving, and then, of course, this process is not so active.
I also use several packages of models in my EAs, trained on different time periods of history, for example from 2007 to 2020 and test from 2020 to today or trained from 2020 to today and test from 2018 to 2020, the latter have shown the best so far.
Now in my advisors 5 suchpackages of models with different periods of training on 28 pairs, the period everywhere hourly.
Then I trade the best pairs in terms of profit, and the best periods for training, for example, the best 3 first pairs out of 28 pairs, and then after such a strict selection with improvements of models, the choice of periods for training, using 28 pairs, trading is just perfect, even in times like these, when markets behave unusual for history.
By the way, the robot is strangely very good at showing the beginning of financial crises throughout history, but it is not strange. Now is clearly one of such crises and it seems to be just the beginning.
Hi, yes about the signs it would be interesting to read the article, if you find the time and willingness, very much looking forward to it.
I tried different indicators as signs, I stopped on Bollinger Bands, but strangely enough with settings that make this indicator the same MA indicator that you use. I wanted to roll back to your code, but strangely enough Bollinger Bands showed better results, perhaps because I pass to the training data on 3 lines of the indicator, although they are repeated.
I use 28 trading pairs, and the most important thing is that every week I retrain the models on new historical data, I run all 28 pairs with new models in the strategy tester, and if a model has shown better performance on one of the 28 pairs, I replace it in the Expert Advisor with a better one. I have automated this process, so it doesn't take much time. As a result, every week, especially in the first two months, the models are actively improving, and then, of course, this process is not so active.
I also use several packages of models in my Expert Advisors, trained on different time periods of history, for example from 2007 to 2020 and test from 2020 to today or trained from 2020 to today and test from 2018 to 2020, the latter have shown the best so far.
Now in my EAs 5 suchpackages of models with different periods of training on 28 pairs, the period everywhere hourly.
Then I trade the best pairs in terms of profit and the best periods for training, for example the best 3 first pairs out of 28 pairs, and then after such a strict selection with improvements of models, choosing periods for training, using 28 pairs, trading is just perfect, even in times like these, when markets behave unusual for history.
By the way, the robot is strangely very good at showing the beginning of financial crises throughout history, but it's not strange either. Now is clearly one of such crises and it seems to be just the beginning.
Hi, yes about the signs it would be interesting to read the article, if you find the time and willingness, very much looking forward to it.
I tried different indicators as signs, I stopped on Bollinger Bands, but strangely enough with settings that make this indicator the same MA indicator that you use. I wanted to roll back to your code, but strangely enough Bollinger Bands showed better results, perhaps because I pass to the training data on 3 lines of the indicator, although they are repeated.
Sorry to interfere in your non-childish conversation. Would itbe possible to use Elliot waves signs in MO?
Sorry to interfere in your non-childish conversation. Could the signs of Elliot waves be used in the MO?
Plato's philosophy would be better
You're not very good with Plato. Do you even know Elliot?