a trading strategy based on Elliott Wave Theory - page 80

 
Я так понимаю что картинка была выложенна Вами для того чтобы другие участники ветки могли сравнить свои результаты с Вашими, так как Вы по моему мнение дальше всех продвинулись в построении стратегии. Так вот пытаясь сравнить свой расчет с Вашим мне достаточно было увидеть эти кналы целиком или узнать их период, точнее меня волновал только самый большой так как он в основном и определяет вероятность по крайней мере у меня. Естественно на такой картинке канал с большим периодом около 100 дней целиком ни как не поместится, а отсюда моя просьба и родилась. И отличие от Вашей картинки в 10-15% позволило бы мне предположить что я двигаюсь в верном направлении и не упускаю важных деталей. Ну нет так нет. :(

Okay. It's no problem for me to punch in the bar number in the comments. Here's a link to pictures, have a look, compare, if it can really help https://c.mql5.com/mql4/forum/2006/07/10_07_2006_additional.zip
And so in my opinion everyone is moving in the same right direction. It's just really hard enough to go in different directions implementing the same rule.


Thanks a lot, now I'm completely satisfied :), I was beginning to think, maybe I've done something criminal.
 
... But the utility doesn't allow you to change anything. The utility didn't help me at all. ...


Dear Solandr, the tester leaves the file locked until you run it on another character/period or close the terminal altogether. Maybe you tried it on a newly created/used file?
 
Jhonny:
Like Solandr, I can't understand how Rosh does such fast calculations.

I don't know exactly how Rosh does it. But after his... er... I took a pen and paper and made sure it is really so. The second point for me (as for Rosh- I don't know) consists in using recursion. Like Rosh, in principle I can send explanations by post.
 
Jhonny:
Да я тоже как и Solandr не могу понять как у Вас Rosh происходят такие быстрые вычисления

I don't know exactly how Rosh does it. But after his ... er ... hints that you can count faster :) I took a pen and paper and made sure it's really true. The second point for me (as for Rosh- I don't know) consists in using recursion. Like Rosh, in principle I can send explanations by post.


It's not exactly recursion - accountants call it cumulative total :)
 
Dear Solandr, the tester leaves the file locked until you run it on another character/period or close the terminal altogether. Maybe you tried it on a newly created/used file?

To check, I made several files in the tester on different symbols on M30 (By opening prices). Closed the terminal. Transferred the fxt files to a machine with a freshly installed (or rather restored from the image) Win2003 Server. Ran the utility. Again it did the same thing - it couldn't open file and didn't allow to edit any parameters. It means that it still lacks something to work properly. The result was identical for each file with a different character.

PS: I've got it sorted now! It all turned out to be a broker. I work with InterbankFX. So the files made in his terminal, the utility can not open. I tried the same for Alpari demo and the file opened, changed and recorded perfectly. But now the question is why do I need the corrected file from Alpari if I am playing with another broker? And the most interesting, what can prescribe in the file InterbankFX, because of what the utility can not open it?

PPS: Well if we didn't manage to correct the file of the tester, I place the results of the tester without the total difference for the whole history, but in principle it is quite enough for evaluation of an EA, but then everyone has calculators ;o) https://c.mql5.com/mql4/forum/2006/07/11_07_2006.zip
Presented are 3 files with different time intervals.
Part1_25_11_2004_to_06_09_2005.gif
Part2_06_09_2005_to_10_07_2006.gif
Part3_10_05_2006_to_10_07_2006.gif
The first 2 files are a story divided in half. Therefore, it is written in the first file that testing was going on till the current time, but in fact, the tester was disconnected after 06.09.2005 and did not perform deals for the rest of time.
The third file is given for the last 2 months. The last 2 months turned out to be very profitable for the Expert Advisor (the profitability is 24.53), since there was not much of a serious trend and therefore many reversals. Of course, to make a correct estimation of the Expert Advisor, one should use the first 2 files where a clear trend was present (profitability 1.89 and 3.63), but not the third file with a small number of deals on the history, which is ideal for the Expert Advisor.
 
Jhonny:
Да я тоже как и Solandr не могу понять как у Вас Rosh происходят такие быстрые вычисления

I don't know exactly how Rosh does it. But after his ... er ... hints that you can count faster :) I took a pen and paper and made sure it's really true. The second point for me (as for Rosh- I don't know) consists in using recursion. Like Rosh, in principle I can send explanations by post.


On the one hand of course it would be interesting to see how you do it, but so far I have a lot of other problem areas that are visible without the speed, I will deal with them and then think about the speed.

ZS I've done a bit of optimisation so far... I've started to memorize intermediate calculations when calculating 2/3 and use them to calculate the entire length, but it only saves me about 12-15% of the time.
 
After one day of watching the multichannels in the indicator (and seeing how they constantly float) I came to a conclusion about the necessity of introducing new criteria. That is, the SCO23>SCO criterion in my implementation is not stable. The most annoying thing is not only that the nearest channels can disappear but they can also appear. Whoever read the tale of "DyDvaChe" as a child will understand. :)
One way is to build nested channels from afar. Of course, the calculation speed will be needed here, but first we must build indicators that calculate various functionalities for the specified sample (not for a single channel!) in the future, as well as in the past. I assume to make one indicator, which by checking the global variable (GlobalVariebles()), will change the calculation algorithm (either RMS, or potential energy, or total energy, or "sum of gradients on parabola" (solandr)).

double GlobalVariableGet( string name) <br / translate="no">
Returns value of existing global variable or 0 in case of error. Call GetLastError() to get the error information.
Parameters:
name - Name of the global variable.

Example:
double v1=GlobalVariableGet("g1");
//---- check the result of the function request
if(GetLastError()!=0) return(false);
//---- continue processing



In general, I don't see a frontal solution, although, I repeat, I haven't studied Murray yet.


By the way, the picture in *.gif format is already available (I hope solandr will give up zips.) Here is the link to this picture - https://c.mql5.com/mql4/forum/2006/07/eurusd60m.gif

 
I have not yet found a way of selecting channels that suits me. And one of the difficulties is that from the very beginning I considered it necessary to include the channels selected on previous bars into the selection. Nevertheless, I will make an EA for the story with what I have. Usually a more specific idea of the task as a whole turns out to be very useful when working on its parts. As for parabolas, I suddenly saw them recently, in large numbers. Actually, anyone who wants to see them can see them :). Naturally there is no certainty that these are THE parabolas.
As for the pictures, I've never had any problems with posting them. True, they were mostly gifs prepared by MT. But recently I've posted two pictures from matlab and no problems either. Maybe there's just a size limit?
 
ZS By the way, the *.gif format picture is already available (I hope solandr will give up zips).

With great pleasure I will give up zips as soon as it becomes possible.
I have already tried it 2 times today. The last one here is "MQL4: Picture for the forum on metaquotes".
The most confusing thing is that after all I can upload ZIPs unhindered!!!? What can be the problem with gif and PNG if they all upload freely. gif pictures prepared in Photoshop 6 as before, but only before they were inserted in that branch, but now in no way!
I took a picture of Rosh specially saved on my computer and tried to put it to cut off my problems with the preparation of images for publication on the site. But the result is zero. Just a real miracle, a simple operation can not be performed and what the problem may be is not clear. Windows2000SP4 Rus system tried from 2 of these computers.
 
ЗЫ Кстати, картинка в формате *.gif уже можно ставить (Надеюсь, solandr откажется от зипов.)

Would be more than happy to give up zips as soon as possible.
Tried it 2 times today already. The last one here is "MQL4: Picture for forum on metaquotes".
The most confusing thing is that after all I can upload ZIPs unhindered!!!? What can be the problem with gif and PNG if they all upload freely. gif pictures prepared in Photoshop 6 as before, but only before they were inserted in that branch, but now in no way!
Especially I have saved Rosh's picture on my computer and tried to put it to cut off my problems with the preparation of images to be published on the site. But the result is zero. Just a real miracle, a simple operation can not be performed and what the problem may be is not clear. Windows2000SP4 Rus system tried from 2 of these computers.


You may say the same problem :(
Reason: