![MQL5 - Language of trade strategies built-in the MetaTrader 5 client terminal](https://c.mql5.com/i/registerlandings/logo-2.png)
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
So you think GA should produce clean results?
This approach is fundamentally wrong. GA should be used to quickly find possible clusters of good solutions and then searched for complete brute force inside a small field of each cluster.
For that you need to run GA (as well as any other, no silver bullet) optimization several times. Random will allow you to jump out of the occasional stuck local extremum.
And there are no such methods that reduce the counting field by orders of magnitude and "don't miss the gold mine". There are no such methods.
Ok, I see your point.
1. How were the "workers" selected from the total number of total enumeration (what are the criteria)?
2. Why didn't the same principles (criteria) as in p1. apply to optimization with GA?
Don't answer me please. Answer these questions for yourself.
You are overreacting. Half of your posts in this thread are filled with insults and personal attitudes.
Not least for this kind of whining and constant flag-pulling of political discontent people went to the ban.
I think the old-timers would agree with what I said. But they will not start a flurry, even though it is Friday.
Bah, it's hrenfx who asked for his account to be removed. It all made sense right away.
...
Bah, it's hrenfx who asked for his account to be deleted. It all made sense at once.
It's a night of miracles tonight!!!
I'm not a hrenfx, as much as you'd like me to be.That's what I would have written ))
ZS: If a soldier says he didn't take it, he'll give it back. hfx is a military man )
zaskok:I'm not a hrenfx, as much as you might like that.
Too bad, but it's still interesting to chat ))
I'm not hrenfx, as much as you might not want me to be.
It's miracle night tonight!!!
And there are no miracles here. I personally don't know if it's him or not. But the fact that he knows about optimization and what it is all about is undeniable. And many traders can be kicked out of the forum, but you can't kick them out of the market, because they are not just good traders, but they also track innovations, changes in trading platforms, looking for new ideas and do not stand still ... And the emergence of new technology in a known platform for them is bound to attract attention.
For example. Let's analyze the tester and the built-in GA optimization. The optimization algorithm is not the only one, they come in different variants that are suitable for solving a certain problem to some extent. There is no single universal method. But what is most valuable from my point of view is that you can involve a large number of computers to solve a task, you can parallelize the task.
After all this tester can be used not only in the way it was meant - search of optimal parameters of some TS. And for example in the interest of the Defense Ministry there are tasks which demand great computational resources. Well for example search of an optimum form of uncertainty function is very important for the task of detecting missile launches by the enemy. At one time I was ready to offer the developers money from the state defense order to perform the required calculations for the Russian Ministry of Defense, and can you believe it is quite big money, but again restrictions .... The functionality is there, it is ready, but I cannot use it because of far-fetched constraints ...
Going back to TS and its optimization, do you know how frustrating it was to sit and write this optimizer myself, when it was ready to use, you just needed to press the button (but it worked only with its history, damn it :-((())
So you sit and think it up, I personally combined two methods, montekarlo and GA. First montecarlo and then GA at interesting points. But how much effort, time and labour it cost me (((.
Therefore, it was reasonable to organize my own under-tester and make a complete enumeration in a dozen or two hours (months in MT). And then in a huge matrix to dig out a workable variant. Or an alternative - a better heuristic algorithm.
Can you give me an example of an algorithm, when in "undershooter" a complete overlap - in a dozen hours, and in MT - months ?
My experience says that all really working TCs are characterized by stability of parameters, i.e. maximum in utility function space is not a "narrow peak" at all, but rather wide area. In a single run of the GA it can of course "miss it" by catching at another extremum, but when the GA is used properly, when prospective areas are first roughly identified and then each individual case is examined, it is impossible to see the area maximum.
Renat, so how is it with debugging in debugger on historical data ? I hope it will be soon ?
...
Renat, how's it going with debugging in debugger on historical data ? Hopefully it will be soon ?
In my opinion, there is a need for MetaQuotes to introduce such a position as an engineer who demonstrates the capabilities of the MT5 platform.
MT5 has gradually turned into a rather complicated complex. Without an explanation and demonstration of MT5 capabilities by the developers, it becomes difficult to cover and use all the features of the platform.
The task of this specialist would be to show on simple examples how to effectively use the capabilities of the platform, to solve trading tasks, to research financial markets, etc.
This is just my opinion.
You don't need engineers, you can't afford the meta-quotes of 1000 engineers, but adequate lessons with a description of how to do it right, how to work properly, how to use this or that feature is very much needed...
A simple example, introduce the OOP, and then what? How to properly use it? Where are the simple lessons like in Kovalev's tutorial?
Maybe instead of another new gadget for mt5 to make a grammatical and adequate training?