Your symbols and your datafeeds in Metatrader 5 - page 15
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
Here's a link to Wikipedia's description of the Annealing Method. https://ru.wikipedia.org/wiki/%C0%EB%E3%EE%F0%E8%F2%EC_%E8%EC%E8%F2%E0%F6%E8%E8_%EE%F2%E6%E8%E3%E0 There is a video example explaining how this algorithm works. It clearly shows that it iteratively approaches the maxima and finds almost the maxima themselves. But it finds other maxima as well. The number of calculations is much less than GA, it is more compact and much closer to the real data than GA and in any case the speed will do the cumbersome and distracted from the real data GA.
Holy naivety, nurtured in synthetic smooth functions.
About significantly fewer passes is complete nonsense. It's enough to imagine a 10 to 13th degree tear-space calculation field and how there "significantly less than 10,000 passes of the regular GA" will be some obviously less universal method to get a laugh out of sheer ignorance of the topic and technical conditions.
The advantage of GA over specialised (based on prior/partial knowledge of the behaviour of the function under study) is that it is more versatile and effective in the gap search mode, which is what trading systems are.
Roughly: in trading systems, usually a miniscule change in parameters completely rips the result, which spoils the life of gradient algorithms hoping for smoothness of the function.
Renat:
The advantage of GA over specialised (based on prior/partial knowledge of the behaviour of the function under study) is that it is more versatile and efficient in the rip search mode, which is what trading systems are.
Roughly speaking: in trading systems, usually a minor change in parameters completely tears the result, which spoils the life of gradient algorithms that rely on the smoothness of the function.
Heuristic trading systems (the optimizer was created for them, in the first place) should be good at finding "smooth" local extrema in the ravaged space of TS.
Holy naivety, nurtured on synthetic smooth functions.
About significantly fewer passes is complete nonsense. It's enough to imagine a 10 to 13th degree tear-space calculation field and how there "significantly less than 10,000 passes of the regular GA" will be some obviously less universal method to get a laugh out of sheer ignorance of the topic and technical conditions.
The advantage of GA over specialised (based on prior/partial knowledge of the behaviour of the function under study) is that it is more versatile and effective in the gap search mode, which is what trading systems are.
Roughly: in trading systems, usually a miniscule change in parameters tears the result completely, which spoils the life of gradient algorithms that rely on smoothness of the function.
If you support the idea of "finding in a huge rip-off can be significantly faster and more accurate than our GA", then yes.
But you are not using Metatrader 5 GA, so you have no way to compare.
If you support the idea of "finding in a huge rip-off can be significantly faster and more accurate than our GA", then yes.
But you are not using Metatrader 5 GA, so you have no way to compare.
When I was doing this for myself, I had already been poking around in heuristic algorithms for quite some time and understood their differences and positive and negative sides and paid special attention to the gradient, as I was trying to make even faster aglorithms with logical search. In my version I made it so that the algorithm would find the group of best data 5-10-20 and the stopping criterion would be the group, rather than a single maximum, to avoid getting to a local extremum. As for universality, it works equally well with 10000 data and with 10 to the -eenth power. That is, the generality of the algorithm does not suffer much, but there is a slight negative quality of approaching the local extremum quite quickly and sifting out nearby data. The GA algorithm in MT produces a wider range of data. But I also had a second optimization option done for extended approximations, in the inclusion file that I have attached is the second function.
Your original message was that the algorithm is significantly faster than GA.
I pointed out that "significantly faster than 10,000 passes in a wild calculation field is a ridiculous statement". Not in a field of 10000, but "regular GA is optimized for speed and allows for 10 000 - 12 000 passes to find solutions in a virtually unlimited search space".
Apparently you don't understand this meaning and don't have comparative figures to present to the public to reproduce the results. That's why stories instead of figures go, and even references to wikipedia without understanding the essence of the process.
I don't use GA MT5 because I can't upload my brokers data and quotes with ascii there, plus I consider MT5 tester as the first, not very successful for users, attempt to improve the tester. It's not user friendly and is made as if by programmers, not traders. Isn't the GA algorithm itself in MT5 different from MT4. That is, I don't mean paralleling the calculation and using CPU cores.
That's where you should have started - you don't use the tool, you have a superficial impression of heuristics, but you make amazing claims.
Made by non-traders is an apotheosis of previous statements. For a start use what has been created - in MT5 the whole tester is qualitatively different, including the analytics system.
We have decided to open the interfaces for writing your own datafeeds for MT5.
So, anyone who wants to not only babble, but also provide their algorithms for testing and comparative analysis, which would once and for all close the subject of "which algorithm is better?
You do not need to open the source code of the algorithm, it is sufficient to provide the compiled core of the algorithm and the inludes (to eliminate possible cheating), which will display the calls of the algorithm and prescribed fitness function itself.
A special welcome to those who criticize MT, be my guest.
As soon as more than 3 people including me want it, we can open separate thread for testing purposes. That in the future to poke everyone in this thread, and that professor including, who "never saw acceptable results".
PS. Thanks to everyone who directly or indirectly helped develop the algorithm.
I have no doubt that your algorithm will overtake everyone.
But for the sake of the experiment, I'm willing to be a third )
I can experiment with standard GA, I already showed one test here.
I also have my own light GA that I developed for easy integration in Expert Advisors. It does not pretend to be fast, but I may try it for the sake of an experiment.
Give me a test function.
That's where you should have started - you don't use the tool, you have a superficial impression of heuristics, but you make amazing claims.
Made by non-traders is an apotheosis of previous statements.