Discussion of article "Brute force approach to patterns search (Part III): New horizons" - page 3
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
Cool) especially cool looks the ratio of maximum loss to total profit, just a huge space for increasing the profitability of the whole system as a whole.
Too bad there are not enough trades...In this thread, systems are not finalised, but new patterns are found and added to the portfolio of systems ;)
This thread is not about refining systems, it is about finding new patterns and adding to the portfolio of systems ;)
First of all you need to understand in this article is already outdated version of the programme, it is already much stronger, you just need to give it a pretty good time and patience. You should also take into account that I have only one working computer with 4 cores, quite old, well plus one more person brutes also helps, but it all happens in his spare time and so far on the record numbers no sense to count on, but soon everyone will be able to participate in it. What 2 people without power can do even with software ?) They can do some things, but it's a small thing. I'll open the veil of secrecy a little bit. The goal is that everyone can take this software and contribute to finding working settings. This can be done by everyone without exception as the programme saves the user from having to program. Using this software is equivalent to ordering it from a marketplace, with the order being "make me a working algorithm of your choice". When simply ordering in the market nobody will give guarantees. Here it is different. The chance of working in the future will be higher than the average on the market, in any case it is already so, even at the testing stage. The main goal is accessibility for everyone. I really want everyone to understand what all this is done for.
Do you think the 2nd option is working?
direction of the transaction after a period of time. As an example of owls in the trailer.
First of all you need to understand in this article is already outdated version of the programme, it is already much stronger, just need to give it a pretty good time and patience. It should also be considered that I have only one working computer with 4 cores, quite old well plus one more person brutes also helps, but it all happens in his spare time and so far on the record is no point to count, but soon everyone will be able to participate in it. What 2 people without power can do even with software ?) They can do some things, but it's a small thing. I'll open the veil of secrecy a little bit. The goal is that everyone can take this software and contribute to finding working settings. This can be done by everyone without exception, as the programme saves the user from having to program. Using this software is equivalent to ordering it from a marketplace, with the order being "make me a working algorithm of your choice". When simply ordering in the market nobody will give guarantees. Here it is different. The chance of working in the future will be higher than the average on the market, in any case it is already so, even at the testing stage. The main goal is accessibility for everyone. I really want everyone to understand what all this is done for.
As soon as you start getting "tradable" results, your goal will change significantly :)
Do you think option two works?
If you are talking about those variants that I typed for this article, I do not recommend using any of them. The whole point is that they are demo variants. In general, they are needed to demonstrate that they work on the forward, because without that they are useless. There are a bunch of methods that can give similar things in an unfair way. I used to make such owls for fun, but now I am not interested in fun. In addition, there will be some techniques that will save even the uninformed from loss.
If you are talking about those variants that I have typed for this article, I do not recommend using any of them. The whole point is that these variants are demo variants. In general, they are needed to demonstrate that they work on the forward, because without that they are useless. There are a bunch of methods that can give similar things in an unfair way. I used to make such owls for fun, but now I am not interested in fun. In addition, there will be some techniques that will save even the uninformed from loss.
I'm not talking about the article, look at the code in the comment at the link.
I don't mean the article, look at the code in the comment at the link
It may work, first of all I don't know what method is used there. The only thing I can say is that of course a data matrix is better than just an array of a particular tool, but with this approach you face the fact that you have to input several data arrays to fill the matrix. A single array won't get everything going. I use purely one instrument data array and analyse it. It does not give super profits, but it simplifies the analysis and you do not need data from neighbouring instruments. All variants work really, just depending on what you want to get in the end. It's all about profit/time ratio. Machine learning is only in its infancy. There are many approaches. I can also make such graphs, draw vector fields, clump results and so on. You can put such mathematics there that no one will ever understand what is the point, what are the main predictions and how long it will all work. How many systems of such systems in a unit of time can be riveted? Can ordinary users do it? I think the answer is obvious to everyone. Anything can work, but it is better to develop your own approach my advice to you.
It may work, but first of all I don't know what method is used there. The only thing I can say is that of course a data matrix is better than just an array of a particular tool, but with this approach you are faced with the fact that you have to input several data arrays at once to fill the matrix. A single array won't get everything going. I use purely one instrument data array and analyse it. It does not give super profits, but it simplifies the analysis and you do not need data from neighbouring instruments. All variants work really, just depending on what you want to get in the end. It's all about profit/time ratio. Machine learning is only in its infancy. There are many approaches. I can also make such graphs, draw vector fields, clump results and so on. You can put such mathematics there that no one will ever understand what is the point, what are the main predictions and how long it will all work. How many systems of such systems in a unit of time can be riveted? Can ordinary users do it? I think the answer is obvious to everyone. Anything can work, but it is better to develop your own approach my advice to you.
Are you referring to #23105? There is a sample EA in the trailer. The idea is the same as yours - bruteforce, but completely random, without trying to analyse anything.
You mean #23105? There's a sample EA in the trailer. The idea is the same as yours - bruteforce, but completely random, without trying to analyse anything.
Bruteforce is the maximum number of runs per unit of time, based on some model. The model could be anything. I have one, another person has a different one. There can be countless models. Most models can be made to work one way or another, regardless of the model, unless of course it is made up in an inept way. But to do such a thing inside the terminal in MQL is only to heat up your iron for nothing. The results will be of course, but it will be faster on additional software. I looked at the code of the mqh file, but expectedly I did not understand anything. If you really need advice and opinion, you need to explain what the method is based on and what you want from it. I just don't have time to study the whole thread and read what someone wrote there and why and solve puzzles. In a private message I'll see what you have there
Bruteforce is the maximum number of runs per unit of time, based on some model. The model could be anything. I have one, another person has a different one. There can be countless models. Most models can be made to work one way or another, regardless of the model, unless of course it is made up in an inept way. But to do such a thing inside the terminal in MQL is only to heat up your iron for nothing. The results will be of course, but it will be faster on additional software. I looked at the code of the mqh file, but expectedly I did not understand anything. If you really need advice and opinion, you need to explain what the method is based on and what you want from it. I just don't have time to study the whole thread and read what someone wrote there and why and solve puzzles. In a private message I'll see what you have there
You don't need to read the whole thread, the whole point is in that post. Completely random bruteforce. 1 variant "optimise" sid pgcch, another variant change sign through N bars, N is selected by optimiser. Since you have not done this I'll try to transfer to opencl, mt does not allow to test it normally.