MT5 terminal updated today and the "Optimisation" window does not show up during the test - page 21

New comment
 

Service Desk has admitted to a hang-up error of forward optimisation.

They said they would fix it.

Waiting.

 
Сергей Таболин:

I notice that trades are marked in different colours.

Maybe it would be better to colour the profitable and losing trades to make them clearer.


No, we shouldn't. It is clear from the vector and colour whether it is profitable or not.

 
Сергей Таболин:

No, what the ***???

Why the fuck would anyone decide that with this ratio, optimization should be interrupted?

Please watch your vocabulary.

 
Сергей Таболин:

I'm sorry.

Well, what else would you call it?

It's like that joke: "Well, then OH... ** **** **** "

 
Сергей Таболин:

No, what the ***???

Why the hell would anyone decide that optimization should be aborted at this ratio?

I pointed out to you that games with massive INIT_PARAMETERS_INCORRECT output in genetics are dangerous.

You just kill the very idea of genetic selection, give out 502 failures out of 512 attempts and don't evenknow what you are doing. Do you want to continue genetics by leaving 10 individuals to breed???

Read articles about genetic algorithms and stop using INIT_PARAMETERS_INCORRECT in genetics:

 
Сергей Таболин:

So what do you want to do?

I recommend it for genetics:

  1. Read up on the mechanics of genetic enumeration
  2. Choose a sufficiently smooth target function to allow the algorithm to see improvements in generations and move forward. Don't fool the algorithm with false results. The genetic algorithm is a rather delicate and sensitive thing.
  3. Don't kill more than a quarter of the population per pass through INIT_PARAMETERS_INCORRECT (use in critical cases, or better not use at all). A quarter is a rough estimate and by no means a recommendation or permission to do so.
  4. The parameters to be run must be operational over the whole interval. Do not knowingly make holes in searched parameters, so that you don't generate INIT_PARAMETERS_INCORRECT. You are doing this solely for your convenience, not to think about the process of brute force
Yes, you need to rearrange the parameter logic to suit the brute force process and the search method used, not your own convenience.
 
Renat Fatkhullin:

I recommend it for genetics:

  1. Read about the mechanics of genetic enumeration
  2. Choose a sufficiently smooth target function to allow the algorithm to see improvements in generations and move forward. Don't fool the algorithm with false results. The genetic algorithm is a rather delicate and sensitive thing.
  3. Do not kill more than a quarter of the population per pass through INIT_PARAMETERS_INCORRECT (use in critical cases, or better not use at all). A quarter is a rough estimate and by no means a recommendation or permission to do so.
  4. The parameters to be run must be working for the whole interval. Do not knowingly make holes in searched parameters, so that you don't generate INIT_PARAMETERS_INCORRECT. You are doing this solely for your own convenience, and do not think about the process of brute force.
Yes, you need to rearrange the parameter logic to suit the brute force process and the search method used, not your convenience.

Doesn't genetics need variable logic? I mean, can we take 100 variables (with a different range of values) and make 4 of them and give it to genetics? Yes, we could, but the principle would be violated, we would get total randomness and non-functional genetics. So how do you propose to do without INIT_PARAMETERS_INCORRECT, when the variable check is used just to exclude illogical values of variables in conjunction with other values of variables or to exclude enumeration ranges? I'm really curious, you can do a lot of manual work to number values by excluding variables, but it's not efficient! How to do it efficiently?

 
Aleksey Vyazmikin:

Doesn't genetics need variable logic? I mean, can we take 100 variables (with a different range of values) and make 4 of them and give it to genetics? Yes, we can, but the principle would be violated, we would get complete randomness and non-functional genetics. So how do you propose to do without INIT_PARAMETERS_INCORRECT, when the variable check is used just to exclude illogical values of variables in conjunction with other values of variables or to exclude enumeration ranges? I'm really curious, you can do a lot of manual work to number values by excluding variables, but it's not efficient! How do you make it efficient?

I don't know where you got your question from.

I said sensibly and deliberately choose your parameters and don't use INIT_PARAMETERS_INCORRECT when you've picked up a very delicate instrument. You should wash your hands before using a microscope, not fight for the right to do whatever you want when you walk into a sterile environment. The power of genetics cannot be ruined by a careless attitude - you just won't get the results you want. Do you think the magic of reducing search area by NN orders of magnitude is given for nothing? On the same conditions as a total messy overshoot?

A complete misunderstanding of the principles has been demonstrated above. Kill 502 individuals in a 512 population and then blame the colony for being dead and not wanting to breed further.

What and how you choose as parameters is your own business. But my business is to point out the minimum purity of experiments if you make deliberate mistakes.

Read the theory, at last.

 
Сергей Таболин:

There's nothing you can do to reconfigure it. It will take months of work to manually prescribe all the permissible parameters. And you have to figure out how to pass it all to the optimizer. It's easier to kill yourself.

But if the first population has given zero result, how hard is it to build a new one? Of course, it's easier to cut off optimization and say that there is nothing to reproduce, than to account for non-working parameters and rebuild a population into a workable one...

You shouldn't have removed the most important thing: do you think that the magic of reducing search area by NN orders of magnitude is given for nothing? On the same terms as a complete dirty overshoot?

I recommend learning and listening to those who spend their time helping you.

Especially when it's one of the direct developers.

 
Renat Fatkhullin:

I don't know where you got your question from.

I said choose parameters wisely and deliberately and don't use INIT_PARAMETERS_INCORRECT when you have picked up a very delicate tool. Hands should be washed before using a microscope, and for fighting for the right to do whatever you want when you walk into a sterile environment. The power of genetics cannot be ruined by a careless attitude - you just won't get the result you want. Do you think the magic of reducing search area by NN orders of magnitude is given for nothing? On the same conditions as a total messy overshoot?

A complete misunderstanding of the principles was demonstrated above. Kill 502 individuals in a 512 population and then blame the colony for being dead and not wanting to breed further.

What and how you choose as parameters is your own business. But my business is to point out that a minimum of purity of experimentation should be observed if you make deliberate mistakes.

Read the theory, at last.

The point is that INIT_PARAMETERS_INCORRECT is often used to narrow down the search area by eliminating illogical/unnecessary values, which significantly speeds up optimization when using full brute force. If INIT_PARAMETERS_INCORRECT didn't exist, genetics would just run into unnecessary variants and fiddle with them, so not excluding variable values is not an option. So that's what I'm asking, how do you suggest, as a professional, to solve this problem without too much effort. To enumerate all correct variables manually, for enumeration without gaps and knowingly exclude INIT_PARAMETERS_INCORRECT values is not an option - very long. Using smart counters is not for everyone and it may also harm genetics.

So what to do, make a pass on all parameters on 1 day and write their correct values in code, assign an ordinal number, and thus select them for genetics?

1...141516171819202122232425
New comment