a trading strategy based on Elliott Wave Theory - page 73

 
.... About the levels - apparently a lot of them are just clusters of stops, i.e. subject to the laws of psychology. Since people behave to a high degree in the same way... <br / translate="no">


As far as I know quantum mechanics does not understand physics of processes, just laws of probability are used there (i.e. we don't give a damn about psychological or climatic or geothermal reasons, the main thing is that cumulative influence of many factors leads to this result) that also has close relation to a theme of this branch. But never mind that later...
 
Yes, a good Occam's razor on the big guys would be nice :)
 
Interestingly, while debugging the script, a channel (or rather, a matching series of 8 channels) evaporated. In this script the channel search goes on for 3000 bars, the first matching channel was around 238 bars and now it just evaporated. Specially checked the parameters via output to file, no error. That's how softly the short channel has evaporated, but the long one has remained. I searched on other timeframes and found other channels there. It turned out to be not so easy, the picture was floating...
There were two episodes, there was one left, as if a hollow suddenly tightened.
 
Candid 06.07.06 15:49

Well, now we need to decide how to deal with close(fan) channels and we can sit down for a tester advisor. Alas, I don't have the time measurement procedure below a second handy.


Candid, by how much length in depth did you calculate you got 3 channels? And at what prices? It turns out that getting more than two channels (channel series) is not so easy.

[/quote]
 
Rosh:
Candid, by how much length in depth did you calculate you got 3 channels? And at what prices? Turns out getting more than two channels (channel series) is not so easy.
I did a run for 10,000 bars on the watch and stopped understanding something.

I counted for the standard 1,000 bars, one pass. Prices so far Close. We can assume it's about the RMS(2/3) > RMS criterion. I'm not using it yet - I don't want to exclude broken channels from consideration. However, it turns out that I am also missing noteworthy channels.


2006.07.08 07:50:26 VGChannels EURUSD,M30: Series time: 0 mc 2006.07.08 07:50:26 VGChannels EURUSD,M30: Channels in the series: 1000, 6 selected 2006.07.08 07:50:26 VGChannels EURUSD,M30: initialized

There are actually 3 channels here, because I haven't decided what to do with the series yet.


 
<br / translate="no">Rosh 08.07.06 00:06

Candid, what length in depth did you count to get 3 channels? And at what prices? It turns out that getting more than two channels (channel series) is not so easy.
I did a run for 10,000 bars on the watch and stopped understanding something.


I think it's about the currency pair try on the pound or the franc, and just at this point in time there are really few channels (although probably ok :)) two groups on the euro one group on the pound
 
P.S. My bad, I lied. For a thousand channels in 1 pass the calculation depth is 3042 bar.
 
You misunderstood me. I have calculated (10000-45)=9955 channels. Channels that meet the criterion usually go one after the other, forming a series. This series is followed by another series of channels that do not meet the criterion. Then the next vadid series and so on. So on the 10,000-bar watch, 4 series were found, in each series the best channel was selected, numbered and the first bar of the channel was indicated. Only the first channel is drawn.
Just in case, I am attaching a 110kb file https://c.mql5.com/mql4/forum/2006/07/Channell5EURUSD30.zip
 
Rosh 08.07.06 13:44

I figured you were interested in the fact that it's like I get more channels (channel series). I would have to specially add code (not only Prints:) to give boundaries of "series" (in your terminology), because series are not explicitly detected in that code. The "good" channels come out of the general loop at once, some of them, as you can see from the picture, may belong to the same series. In the near future, with the implementation of the almost 'finalised' new way of selecting channels, I will try to post suitable figures for comparison. As for drawing the channels, I find that looking backwards allows a better understanding of the history :), accordingly I don't keep detailed logs, limiting myself to looking at the graph. When I'm done with history, I'll turn them to the future too :).
Below is the log, which could be prepared by adding only Print:
2006.07.08 15:28:53 VGChannels EURUSD,M30: Time to compute series: 0 mc 2006.07.08 15:28:53 VGChannels EURUSD,M30: Channels in series: 1000, Selected 6 2006.07.08 15:28:53 VGChannels EURUSD,M30: Channel Number: 6 First Bar: 99 Last Bar: 33 A = 0.00009638 B = -0.9658 R.M.S = 0.0008 2006.07.08 15:28:53 VGChannels EURUSD,M30: Channel Number: 5 First Bar: 105 Last Bar: 35 A = 0.00009039 B = -0.8264 R.M.S = 0.0008 2006.07.08 15:28:53 VGChannels EURUSD,M30: Channel Number: 4 First Bar: 2124 Last Bar: 708 A = -0.00001154 B = 1.5294 R.M.S = 0.0088 2006.07.08 15:28:53 VGChannels EURUSD,M30: Channel Number: 3 First Bar: 2352 Last Bar: 784 A = -0.00000016 B = 1.2792 R.M.S = 0.0102 2006.07.08 15:28:53 VGChannels EURUSD,M30: Channel Number: 2 First Bar: 2382 Last Bar: 794 A = -0.00000112 B = 1.2511 R.M.S = 0.0102 2006.07.08 15:28:53 VGChannels EURUSD,M30: Channel Number: 1 First Bar: 2391 Last Bar: 797 A = 0.00000150 B = 1.2429 R.M.S = 0.0102 2006.07.08 15:28:53 VGChannels EURUSD,M30: initialized



 
Finally made a drawing of several channels.
Especially for Candid, I made one on 30 minutes.