MetaTrader 5 Strategy Tester: bugs, bugs, suggestions for improvement - page 25

 
fxsaber:

It plays when I am in the Graph or Backtest tabs.

Got it. Corrected
 
fxsaber:

Adviser


Playback


Corrected
 

Is it possible to parallelise the testing of the EA?


I have a period of 4 years, 27 characters. EA runs it on one processor in 1.5-2 days. Now I have the following inputs:

1. because of a very large deposit for the 4th year, the Expert Advisor has to trade sets of 500 lots each, which slows down the testing and distorts, taking into account the limitation of maximum number of orders at the broker.

2. I don't need to know the overall final result at the stage of these runs, I need to understand how the EA behaves at particular intervals. I would be happy to divide the whole interval by the number of processors + crossovers (see below) and run pairs of intervals with crossovers. Here's an example: let's say we have 4 processor cores, we get 6 intervals. Then this is how the core runs are taken:

I. 1-3, II. 2-4, III. 3-5 IV. 4-6

Then it would greatly speed up the testing of certain sets. Now I have to run several instances and parallelize them manually. I'm about to run it on 15 years and I'm getting worried).

 

Now the optimisation range data of the input parameters for those that have been disabled from optimisation does not get into the opt-file.

For example, optimised by Range2 and Range1 was not optimised, but the optimisation range was set. So this range does not get into the opt file.

Please add.

 
In the optimisation opt file for all characters from Market Watch the following field is zero
initial_deposit = 0.0

Please correct.

 
Another question about the tester is this. Is there a histogram of income distribution by testing time? Or how do you make one for the runs? Or how do you make the optimisation tends to the average income distribution?
 
Andrey Pogoreltsev:
Here's another question about the tester. Is there a histogram of income distribution by test time? Or how to make it during runs? Or how to make optimization tends to the average income distribution?

Calculate for yourself, return from OnTester the evenness of distribution multiplied by the profit or other criterion of interest.

 
Andrey Khatimlianskii:

Do the math yourself, return from OnTester the evenness of distribution multiplied by the profit or other criterion of interest.

Oh, right! Thanks

 
Slava:
I see. Corrected

Thank you.

What about the distribution of tasks to agents in Full Optimisation mode?-Are tasks given to currently available agents, or is there some other principle of distribution?

 

If you change the input parameters in the Tester in the Settings tab via CTRL+V without going to the Settings tab and start Optimization. Then the old input parameters are placed in the opt file.

To ensure that the opt file receives the correct input parameters, it is mandatory to go to the Settings tab after importing.

Please correct.