The future of MQL5 is MQL5+ or even MQL6 - page 7

New comment
 
Slawa:

I'll let you in on a big secret. At maximum test speed (32) there is no lag. The vice maximum speed (31) has ::Sleep(0). Simply switching the threads gives this difference

But I don't want to use empty loops for delay - other users will get angry: "Why the fuck did I get a 100% CPU load on nothing!"

If Sleep() is not called on every tick?
 
Dmitry Fedoseev:
If Sleep() is not called on every tick?
There is an obvious imbalance in the visualisation
 
MT4:
  1. When selectingthe 'Genetic Algorithm' when optimizing, I would like to specify the minimum number of 'Total Trades' in order to avoid a cycle in which the optimizer chooses the variant with less and less 'Total Trades' while selecting the largest value of the 'Optimizable Parameter'. And the ability to automatically calculate the value of "minimum number of 'Total Trades' " depending on the number of days used during optimization.
    For example: Optimization is done from 2015.01.01 till 2015.12.31 = 259 days of optimization with coefficient "1" equal to 259 (minimum amount of "Total trades") or with "0.5" coefficient equal to ~129 (minimum amount of "Total trades").
  2. Ability to test in descending order of days used for testing (example: from 2015.01.01 to 2015.12.31, next step from 2015.01.02 to 2015.12.31, and so on) in the name of identifying places where the Expert Advisor passes the test by successfully matching the start of testing, 'sitting out' the 'Free Margin' drawdown at the expense of the increased balance from previous trading when the Expert Advisor successfully matched entry and exit. Or the ability to test with initial deposit only, without using the Balance.
 
A counterpart toIntelliSense needs to be introduced.
 
lilita bogachkova:
MT4:
  1. When selecting the'Genetic Algorithm' during optimization, I want to specify the minimum number of 'Total trades' to avoid the cycle in which the optimizer selects less and less 'Total trades' while selecting the largest value of the 'Optimizable parameter'. And the ability to automatically calculate the value of "minimum number of 'Total trades'" depending on the number of days used in the optimization.
    For example: Optimization is done from 2015.01.01 till 2015.12.31 = 259 days of optimization with coefficient "1" equal to 259 (minimum amount of "Total trades") or with "0.5" coefficient equal to ~129 (minimum amount of "Total trades").
  2. Ability to test in descending order of days used for testing (example: from 2015.01.01 to 2015.12.31, next step from 2015.01.02 to 2015.12.31, and so on) in the name of identifying places where the Expert Advisor passes the test by successfully matching the start of testing, "sitting out" the drawdown of 'Free Margin' due to increased balance from previous trading when the Expert Advisor successfully matched entry and exit. Or the ability to conduct testing with initial deposit only, without using the Balance.

+1

All current criteria for the genetic optimizer are 'unrobust', i.e. the result is too overoptimized and unprofitable in the fronttest. In EAs which are available to me I am solving such problems with my own code, checking the minimum number of deals per week etc. In Expert Advisors from the Market I simply do the fullest possible test for all parameters and then process the results in excel.

Many problems would be solved if we could add the possibility to write our own code for ontester() of any EA. It probably will not have access to the EA's global variables, of course, but all data from TesterStatistics() should be readable.


Finished it later:
I thought so, it would be even better if the scripts could call optimisation. Call parameters - date, EA name and its own parameters, etc. Full functionality of the usual strategy tester. At the end of the test, the script could get all the results with full access to TesterStatistics() of each.

 
agvozdezkiy:

2. Make normal versions for Mac and Lin, so that there are no vinyls. Works in it every once in a while.

How many percent are trading under them? 1% or 1.5% ? No need to spread it around.

3. Make it possible not only to "fix" EAs with indicators, but also to update the interface.

I think the implementation of the last point will speed up the development of MT)).

What interface should I update for my trader? Please write clearer.

 
Slawa:

I'll let you in on a big secret. At maximum test speed (32) there is no lag. The vice maximum speed (31) has ::Sleep(0). Simply switching the threads makes the difference

But I don't want to use empty loops for delay - other users will get angry: "Why the fuck did I get a 100% CPU load on nothing!"

Cool! I never would have thought there could be such a difference.
 
Alexey Volchanskiy:
well occam's razor, we just wash)
 
Dmitry Fedoseev:
If Sleep() is not called on every tick?
Or play with thread priorities. Priority values are not much, but as an option for experimentation. The more so, it can be checked in 5 minutes. Hey, an idea occurred to me now to decrease the priority of the terminal itself for the period of a visual test ))))).
12345678910111213
New comment