Errors, bugs, questions - page 458

 
Interesting:
And to pass parameters to another program and work with a separate copy of the indicator there is an option?

Yes. A normal solution.

(There may not be a separate copy if it is an inbuilt indicator. It may just increase the global count of references, if there is already an indicator with these parameters).

 
Interesting:
If by their own ticks, you could try it virtually. But given that it is a multicurrency system, it is unlikely to inspire anyone to implement it (especially if you have to reproduce all trading processes virtually).
There are EAs whose reaction depends on tick characteristics at the moment. And I think (I'm a dilettante in this field ))) that these are system characteristics, and not DC's spike. And testing on virtual ticks obtained from one-minute bars, as it is usually done in the MT5 tester, is not informative enough for periodic parameter optimization in a constantly changing market...
 
rrr:
There are EAs whose reaction depends on the characteristics of the ticks at the moment. And I think (I am a dilettante in this field ))) that these are system characteristics, and not broker's spike. And testing on virtual ticks obtained from one-minute bars, as it is usually done in the MT5 tester, is not informative enough for periodic parameter optimization in a constantly changing market...
The desire is understandable, but in my opinion "The game won't be worth the candles"...
 
Renat:
Please give me enough code to reproduce the situation (run it in a couple of clicks and see).
Emailed two days ago - any news?
 
marketeer:
Emailed two days ago - any news?

Why publish in private?

If you can't post here, there is a dedicated servicedesk

 
stringo:

Why publish in private?

If you can't post here, there is a dedicated servicedesk

What's wrong with private? It's a normal channel of communication, in between the public and the service Desk. Well, a service-desk is a service-desk.
 
rrr:
There are EAs, the reaction of which depends on tick characteristics at the moment. And I think (I am a dilettante in this field))), that these are system characteristics, and not brokerage companies' tips. And testing on virtual ticks obtained from minute bars, as it is usually done in the MT5 tester, is not informative enough for periodic parameter optimization in a constantly changing market...

First, read the article Algorithm of tick generation in the Strategy Tester of the MetaTrader 5 terminal.

If you still want to collect ticks, you should write your own strategy tester in C++.

There is no other solution, there is no tick history in the Strategy Tester 5 and there will be no data substitution with user data.

But I think that if a strategy is so sensitive to ticks quality, it will have problems when passing to another brokerage company.

Every brokerage company has its own filtering system, that's why ticks are different. The brokerage company tries to cut off high-frequency trading using different methods.

That's why if you manage to write an EA, then for sure after some (not very long period) you will get a ban on auto-trading.

Now think whether it is worth spending time on the EA development and software development for its testing, only to be prohibited from using it after a week of profitable trading.

 
Urain:

First, read the article Algorithm of tick generation in the Strategy Tester of the MetaTrader 5 terminal.

If you still want to collect ticks, you should write your own strategy tester in C++.

There is no other solution, there is no tick history in the Strategy Tester 5 and there will be no data substitution with user data.

But I think that if a strategy is so sensitive to ticks quality, it will have problems when passing to another brokerage company.

Every brokerage company has its own filtering system, that's why ticks are different. The brokerage company tries to cut off high-frequency trading using different methods.

That's why if you will manage to write an EA, most probably after some (not very long period) you will get a ban on auto-trading.

Now think about it, is it worth spending time on the development of an EA, development of software for testing it, only to get a ban on its use after a week of profitable trading?

I have read the recommended article before.

As for the bans, this is the business of each brokerage company. Some brokerage companies apply prohibition on short-term stay in the market for, say, minutes, which is quite acceptable. Further on brokerage companies will resist by any means, requotes, delays in execution, "non-market" ticks, etc. etc., up to simple "non-reaction" to you and "non-payment....". Who cares how much about their reputation. But that is another issue. In addition, these problems arise with manual and fairly profitable trading as well......

Of course a lot depends on the individual filters of brokerage companies, but all this is solvable, because every filter has its advantages and disadvantages, ie two sides of the coin, but to learn and make your own tester, with this problem .... ))) The same as in Expert Advisors you periodically need to adjust the parameters, and in manual trading there is no guarantee of repetition of yesterday's positive results, alas, that's the way the market is (imho).

P.S. This need would not be significant if we were only testing on one pair. And because testing has to take into account the movement of many currency pairs, the discrepancy between virtual and real ticks in the "sum" of different pairs becomes significant....

 

Greetings gentlemen!!! Please advise: I requested the opening time of the latest bar on the smallest timeframe (M1 CopyTime(...)), find out for example that the latest bar on M1 opened 2005.09.07 18:20, can I be sure that there is no later opening on higher timeframes???

Is there a function that would return the latest date regardless of timeframe and server (e.g. if I just copied quotes)???

 
220Volt:

Greetings gentlemen!!! Please advise: I requested the opening time of the latest bar on the smallest timeframe (M1 CopyTime(...)), find out for example that the latest bar on M1 opened 2005.09.07 18:20, can I be sure that there is no later opening on higher timeframes???

Is there a function that would return the latest date regardless of timeframe and server (e.g. if I just copied quotes)???

1. It seems that all TFs are built on the basis of minute bars, so if you correctly identify the last minute bar then there should be no information about all other symbols later.

But we should take into account some points, such as connection to the server (yes or no, if not, for how long), and maybe we should control other things as well.

2. If I understood the question correctly, then regardless of the TF on the symbol will return - SYMBOL_TIME, about the server question is complicated (should work everywhere, but will be tied to the time of a particular server and its data flow).