Errors, bugs, questions - page 792

[Olegs Kucerenko]

muallch:

More on genetics. The tester, for some reason, arrogantly believes that with billions of variants he will find the best one in 1280 runs. And he stops after that! Convincing him to count traditionally at least 10 thousand variants doesn't work... Recompiling etc doesn't help. What a mess!

No, really, what to do? I can't stop it at all.

And I have enough 1280 runs (then it expands to a couple of thousand), but sometimes it starts with 10K....To 1280, then 10K.

[Igor Maslov]

Karlson:
And I have enough 1280 runs (then expands to a couple of thousand), but sometimes starts with 10K....To 1280, then 10K.

Well, that's a matter of opinion. It all depends on the demands.

[Renat Fatkhullin]

muallch:

Karlson 2012.08.02 00:30 # After finishing or by stopping the optimization, the Start-Stop button remains inactive and faded.

Confirmed, there is a bug and it is very repeatable. Besides, optimization results of the Expert Advisor I tested just last night have radically changed - e.g. number of trades with the same parameters has decreased by 4 times. Unfortunately, I have not saved the results of yesterday's runs and I can not track why this is so - I am just stating the fact for now.

This error has already been fixed. A new build will be released on Friday.

[Igor Maslov]

Renat:
This error has already been corrected. A new build is coming out on Friday.

That's very good. I have a suggestion. After the release of the aforementioned build, to "freeze" until the start of the championship the introduction of fundamental functional changes, i.e. those that hypothetically could affect the already well-tuned systems. This does not, of course, include bug fixes. It's uncomfortable to go back to the middle of the work already done.

[Валерий]

Renat:

It is very simple here, there is no error:

1. The margin level chart shows the ratio of margin coverage per trade to total account equity. That is, it is a relative value, not an absolute value in dollars.
   
   
2. The lot is fixed, which gives the margin requirement under the trade fully proportional to the IBM share price. The exchange rate did not fluctuate much between $120 and $133.
   
   
3. Since the balance(equity) is increasing and the margin coverage is almost fixed (the exchange rate has not changed much), the margin * 100.0 ratio\ equity will naturally fall.

You have made two errors in your reasoning:

1. Thought that the margin had something to do with the balance, forgetting that the volume of the transaction was fixed.
   
   
2. Didn't pay attention to the fact that the margin level is shown, not the margin itself (the graph is specifically made to show the load on the account).
   

It's strange to hear this from the chief developer. From the user manual of the terminal:

The margin level is the percentage ratio of the amount of funds available in a given account to the amount of margin (Funds / Margin * 100);

[Olegs Kucerenko]

Valmars:

Strange to hear this from the main developer. From the terminal user manual:

The margin level is the percentage ratio of the amount of funds available in a given account to the amount of margin (Funds / Margin * 100);

On my Eurodollar chart the balance is growing strongly, while the margin is not fluctuating as much (constant say), so there is an increase in the level.

On IBM's chart the balance grows much weaker, while the collateral on the shares grows significantly (there is a very heavy load on the depo, leverage is small and everything like that).

This causes the margin level to fall as the balance sheet grows.

The formula is ... out of the question)))

[Сергей]

int TimeDayActivation ( int Iorder ) //Процедура активации советника в сторого определенное время и день недели
   {
//---
      TimeToStruct ( TimeCurrent (), time );
      if ( Iorder == 1 ) if ( time.day_of_week == _1_day && time.hour == _1_hour && time.min == _1_minute ) return true;

//---
      return false;
   }

this kind of code worked fine in an EA for the Champ...

Now there's something wrong with time.min...

If I set 0 minutes, then one number of deals, if I set 5 minutes, then 4 times less... (not to mention changing day of the week and hours... )and it's all the same, why? it's a shame, different number of trades... although logically it can't be with 1-5 min difference...

what else has been ruined in the language in these 6 months? i dont see any improvement... things have gotten worse on all fronts

P.S. you can tell I have mistakes somewhere... again... six months ago everything was working perfectly... deviations were 1-2 transactions... you could ignore that... the tester's fumbles... but not multiple times... how can you trust such tests at all...

[Anatoli Kazharski]

S4kam:

code like this worked fine in an EA for the Champ...

...

Try to display all the values in the log. What does it show?

[Stanislav Korotky]

How do I run optimisation using agents in the cloud? Cloud is logged in, clouds in 4 pieces write ready. In the context menu the Use -> Mql5 Cloud Network option is enabled. However, when optimization starts, only local agents work and cloud agents write all failed.

[Olegs Kucerenko]

marketeer:
How do I run optimisation using agents in the cloud? Cloud is logged in, clouds in 4 pieces write ready. In the context menu the Use -> Mql5 Cloud Network option is enabled. However, when optimization starts, only local agents work and cloud agents write all failed.

It is not visible in the log file why? Because it seems to be enabled, logged in...