a trading strategy based on Elliott Wave Theory - page 274
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
PS: These two posts can probably be deleted...some kind of glitch
Of course, the tester will not be able to handle the bind. It's not built for it. I imagine that in order to use the read from file in the tester, you also have to work really hard.
However, there is no need to test the mapping in the tester. In the tester ticks are generated after computation cycle is completed, so the tester doesn't care whether the program is counting 7 hours or 7 milliseconds. What can not be said about the user who wants to get the results alive :-) That's why for testing you don't need a bundle at all, you just need to combine it in one Expert Advisor. Fortunately this connection, as well as division, is quite elementary.
But there is another problem which exists for both bundle and transfer file. In both cases, the tester produces results "before the arrival" of the next tick, as if the calculation is instantaneous. This is a trap which makes the tester test pointless and misleading. In reality, the results obtained on a certain data set will only be available to the Expert Advisor after the calculation is completed, i.e. after about 2000 ticks. Or, if by "ticks" we mean minute bars, after 400 ticks, which is not a few.
Will these results still be relevant? What is the range of your forecast, Sergey, and what part of it will be eaten by this lag?
At least the documentation says it can, but I haven't tried it yet.
to Yurixx
Will these results still be relevant ? What is the range of your prediction, Sergey, and how much of it will be eaten up by this lag?
Time to calculate simplified model in MathCAD takes about 10-30 min depending on channel length. It calculates one most probable level, till which the price can go from the current price level for some expected time with variation from 3 hours to 1.5 weeks. Test results of the forecast are quite good.
I was about to do it that way once, and even sketched out a variant of the code. But then I changed my mind, so the code was not thoroughly tested. And it is also intended for solving particular issues. But it may come in handy as a seed for reflections.
int start() { if (NextRead && !FileIsEnding(handle)) { T = FileReadInteger(handle,LONG_VALUE); P = FileReadDouble(handle,DOUBLE_VALUE); //Print(TimeToStr(T)); NextRead = false; } if (Time[1] >= T) { while(Time[1] != T && !FileIsEnding(handle) { T = FileReadInteger(handle,LONG_VALUE); P = FileReadDouble(handle,DOUBLE_VALUE); } if (Time[1] == T) { Price[1] = P; DataCnt++; } NextRead = true; }A small note: such indicator can be tested only in visualisation mode.
There is a nuance here. Suppose, after receiving the result of the tester run, a user will decide to try another tactic. Or they will optimize the parameters that are not relevant to the calculation. If he/she has a file, he/she will do it at once, if he/she does not have a file, it will take him/her a very long time. So with these calculation times, you need to think very carefully about what to write to the file.
It depends on what and how to calculate. But here the user must think. But the warning is certainly very appropriate.
However, if you use the script to call the heavy calculation from the indicator, the new data will stop it or not. Will the new data stop it or not? I just don't have anything implemented in MT yet, but I want to know in advance. :о)
The idea is interesting, but for me, for example, it is not so important, because there are no parameters that can affect the subsequent calculation. Or rather, there are parameters, but they are interconnected and calculation of the value of one implies calculation of the next, i.e. they define each other.