Your symbols and your datafeeds in Metatrader 5 - page 12

New comment
 
Laryx:

Something tells me that this approach could easily be done under GA as well. The amount of work is about the same. The "inner loop" is going through on every pass...

But, most importantly - how much demand would that be ? As I suspect, not everybody uses even simple custom OnTester() function. But this is a very powerful optimization tool.

I feel that I missed something, because it's hard to find the logic of transition from answering your question about getting acceleration in under-tester to applicability of GA. Everyone discussed heuristics yesterday, someone even managed to make up a fat point convincing himself that "everything is clear".

I am not interested in arguments and convictions. It was a pleasure to cross paths with you. The rest is empty.

Even the guys who use custom criteria in MT4 using OnTester are doing a great job in trading robots. There are just few of them express their opinions here. And probably they are right.

 
zaskok:


I'm not interested in arguing and convincing.


it's definitely him ))
 
Laryx:

Something tells me that this approach could easily be done under GA as well. The amount of work is about the same. The "inner loop" is going over on every pass...

You are trying to have a dialogue with him for nothing.

This guy doesn't use MetaTrader 5 at all, he uses only 4. You can see it in his logs - he hasn't run MT5 for years, but he criticizes it.

I already caught him a couple of years ago on "you had a single launch of MT5 many months ago, how do you manage to make assessments and criticisms?". He too denied it, as he is now his next clone.

 
From here, from page two, the topic has been abandoned altogether and the discussion of Genetic Algorithms has begun.
 
barabashkakvn:
From here, from the second page, they went off topic altogether and started discussing Genetic Algorithms.
And all because CopyTick doesn't work!)
 
barabashkakvn:
This is where, from page 2, the topic was abandoned altogether and started discussing Genetic Algorithms.
It is high time to move everything to a proper topic (there are several ready-made ones on GAs).
 
I haven't delved deeply into previous GA posts, but I wanted to share that at one time, because MT4 doesn't allow testing with ascii and MT5 doesn't allow me to insert my quotes, I had to write my own tester using scripts. I tried different Genetic Algorithms, including the one described in my articleGenetic Algorithms are Simple! Some of them I rewrote in mql, and some in the form of DLLs. Then I came across another algorithm called the Annealing method. Made a few variants and was amazed at the speed and accuracy of counting. If МТ GA rarely found data close to the extremums from, the annealing method did not find them in rare cases. As a result, I kept a simplified (without temperature, but with a simple logical linear compression coefficient), but fairly accurate version of the Annealing method. Trillions of a year's worth of data on minutes at opening prices, it took him about 5 minutes to calculate. And that's without paralleling the calculation. It's a good thing the Annealing method, if you haven't tried it, take a look at it.
 

Provide evidence, please.

Take a specific example, describe it for independent reproduction, run it in MT5 and your tester and then post detailed data with conclusions. Without that, your words are unacceptable. This is a technical discussion.

 
Renat:

Provide evidence, please.

Take a specific example, describe it for independent reproduction, run it in MT5 and your tester and then post detailed data with conclusions. Without that, your words are unacceptable. This is a technical discussion.

Yes when I wrote this I was comparing a few different options. I was trying to find extreme data first, and then I connected to annealing method and it found these values, or more precisely a group of values, in about 8-9 out of 10 cases. Then I put a copy of the Expert Advisor in the MT4 tester and it found values from this group in about 1-2 instances and the rest were somewhere in the vicinity. The speed can be approximate, while input and output data is only processed in scripts, the counting itself goes in DLL, through the system and some measures, like getting rid of loops and functions for almost linear calculation, have been implemented inside. That is, it's a speed calculator, which is not objective to compare with the tester. Just approximately by eye it is visible, that it is faster when there is a lot of data, and according to the time counter my version is faster not less than 10 times. It is difficult for me to make a deeper and more reliable research. The only thing, if it can help you, I attach the algorithm itself in the form of mqh-file and to make it clear how it is used, running the script.
Files:
MO.zip  6 kb
 

That is to say, there is no corroboration of your words.

This is to say nothing of the possibility of conventional criticism of your claims on a theoretical level.

1...5678910111213141516171819...26
New comment