It is not certain that frequency will improve the results, but it will reduce the speed of the tests for sure. Timer events are not perfect, there will almost always be an error. Even on an empty timer you can see the error. If there are calculations within the timer function, the margin of error is likely to be higher. Loading the computer with third-party tasks can also affect the timer. There is a chance that not all open/close/modify signals will hit at the same time and on the same ticks.
It's worth mentioning that I made all these observations a couple of years ago on a heavy multicurrency tool that uses almost 30 instruments simultaneously. The results only matched when the timer was off. I assume not much has changed now.
Thank you.
What is the other option apart from the timer essentially ? Cycle?
I know that the timer is not an ideal solution. I know that there are nuances of glitches and disabling the timer.
What is the other option apart from the timer essentially ? Cycle?
I know that the timer is not an ideal solution. I know there are nuances of glitches and disabling the timer.
I'm fine withthe spy option. I use a slightly different approach now, but the essence is the same. Often, if possible, spies immediately send signals to close/open/modify, or some key information for forming signals in the Expert Advisor itself. This parallels the calculations, and the Expert Advisor basically receives only trading functions.
В настоящее время мультивалютных торговых систем, индикаторов и экспертов разработано огромное количество. Тем не менее, до сих пор создатели этого "огромного количества" сталкивались со специфическими для мультивалютных систем трудностями. С выпуском в свет терминала MetaTrader 5 и языка программирования MQL5 появилась возможность реализации...
I'm fine withthe spies option. I now use a slightly different approach, but the essence is the same. Often, if there is a possibility, spies immediately send signals for closing/opening/modifying or some key information for forming signals in the Expert Advisor itself. This parallelizes the calculations, and the Expert Advisor basically has only the trading functions.
Thank you. I read this article once.
Spy is good. But what about the tester?
Yes and chart events are also sometimes slow. Sometimes the whole terminal hangs ( I recently encountered a bug when working with copybuffer ) when copying data, graphics hangs for an indefinite time. Including graph events.
Yeah and schedule events also slow down sometimes.
What about the tester? The spies work on the ticks of the symbol on which they are launched. They send events properly.
What do you mean "chart events also sometimes slow down"? I need examples. For example, I do not send ticks, but send mostly signals - the event queueis not overloaded and there are no brakes.
Vladislav Andruschenko:
Sometimes the whole terminal hangs (I encountered a bug recently with copybuffer) when copying data, graphics hangs for some time. Including graphics events.
Well, the bugs, of course, are related to the subject of the branch, but we are talking about the principle of multicurrency. To analyze the bugs we need examples.
What about the tester? The spies there work on the ticks of the trading instrument on which they are running. They send events in a good way.
What do you mean "chart events also sometimes slow down"? I need examples. For example, I do not send ticks, but send mostly signals - the event queueis not overloaded and there are no brakes.
Well, bugs, of course, are related to the subject of the branch, but we are talking about the principle of multicurrency. For parsing the bugs we need examples.
Thank you. I'll try to parse and check variants with chart events.
OK. The answer is above.
Thank you Constantine for replying. Please don't be too distracted by other things!
It is not certain that frequency will improve the results, but it will reduce the speed of the tests for sure. Timer events are not perfect, there will almost always be an error. Even on an empty timer you can see the error. If there are calculations within the timer function, the margin of error is likely to be higher. Loading the computer with third-party tasks can also affect the timer. There is a chance that not all open/close/modify signals will hit at the same time and on the same ticks.
It's worth mentioning that I made all these observations a couple of years ago on a heavy multicurrency tool that uses almost 30 instruments simultaneously. The results only matched when the timer was off. I assume not much has changed now.
Thank you.
Cycle will not work.
The multicooker is a pain in the ass, there's no good solution.
the cycle will not work.
The multicurrency is a pain in the ass, there is no good solution
on M1 no resemblance is possible in principle, on the others it is possible
the developers made it so that now you can almost any indictor with custom symbols to take it apart.
I don't know whether to be a developer or better to copy
Thank you.
I'm fine withthe spy option. I use a slightly different approach now, but the essence is the same. Often, if possible, spies immediately send signals to close/open/modify, or some key information for forming signals in the Expert Advisor itself. This parallels the calculations, and the Expert Advisor basically receives only trading functions.
I'm fine withthe spies option. I now use a slightly different approach, but the essence is the same. Often, if there is a possibility, spies immediately send signals for closing/opening/modifying or some key information for forming signals in the Expert Advisor itself. This parallelizes the calculations, and the Expert Advisor basically has only the trading functions.
What about the tester? The spies work on the ticks of the symbol on which they are launched. They send events properly.
What do you mean "chart events also sometimes slow down"? I need examples. For example, I do not send ticks, but send mostly signals - the event queueis not overloaded and there are no brakes.
Well, the bugs, of course, are related to the subject of the branch, but we are talking about the principle of multicurrency. To analyze the bugs we need examples.
What about the tester? The spies there work on the ticks of the trading instrument on which they are running. They send events in a good way.
What do you mean "chart events also sometimes slow down"? I need examples. For example, I do not send ticks, but send mostly signals - the event queueis not overloaded and there are no brakes.
Well, bugs, of course, are related to the subject of the branch, but we are talking about the principle of multicurrency. For parsing the bugs we need examples.
So if unconditional pickup of inludes is implemented, then macros + templates won't create a problem with it, right?
Why are you fighting? Why don't you do something useful together?