logo
Welcome Guest! To enable all features please Login or Register.

Notification

Icon
Error

Options
Go to last post Go to first unread
guara_riua  
#1 Posted : Wednesday, October 26, 2005 6:27:48 PM(UTC)
guara_riua

Rank: Advanced Member

Groups: Registered, Registered Users
Joined: 4/24/2005(UTC)
Posts: 77
Location: Canada

Hello everyone I just bought the ADSI package and I am working on "mastering" it the fastest I can. I've built 2 reference value tables, one for Alphas and the second for the DeltaPhaseLows and DeltaPhaseHighs, in order to have the parameter values handy when adjusting Ehlers indicators that require them. However, using real currencies/equities doesn't give me a clear understanding about the output response when changing the indicator param values. I've tried to use a fixed frequency [with Sin(Cum(1)*360/K)] where K is the number of bars/cycle I want to test the indicators input. I've tested various param values with the SineWave that acts like any other filter with poles and zeroes, which is perfectly normal. However, the SineWave output varies pretty much for a given frequency when changing the parameters, which is a normal behavior too. What I am looking for is a stimulus set of files that will have the data values properly generated to test and understand better the adaptive indicators that are part of the ADSI package. I assume Brad might have already this kind of stimulus file set (in a CSV/XLS format) because he had to test all the functions with clean mathematical stimulus (like Sin, saw shape etc.) An example of stimulus files I am looking for are described below: 1) "Chirped" Sine Wave: This would be a Sine wave that varies its cycle from 2->240 bars but can have from 1->12 complete cycles (here can be 12 files) at the same frequency before it will change upward. 2) "Heterodyned" Sine Waves input This would be a low frequency Sine wave (i.e. 80-240 bars) as the dominant cycle multiplied by another Sine Wave with a a smaller amplitude (38-62%) and a higher frequency (8-40 bars). 3) Triangular-Sine complex signal This one would have a linear variation of the price as the dominant cycle (a triangular saw shape input = 80-240 bars) multiplied by a Sine Wave with a smaller amplitude (38-62%) and a higher frequency (8-40 bars). The reason for this stimulus is to test and understand how the parameters listed above(Alphas, Deltas) affect the indicators output: - response time - output wave distortions vs the input wave - the LAG variations with frequency - output frequency "cutoff" or attenuation as function of input frequency - the indicators ability to extract the cycle component from the complex input signal Eventually, will be possible to extract optimized values or simplified formulas to calculate the Alphas and Deltas as a function of the Cyber Cycle i.e. My expertise being in the electronic hardware design, I don't have a programmer proficiency and speed to generate these stimulus files. The code development and debugging will be quite time taxing for me. I would be eventually able to create these files using VHDL language, but any help will be appreciated, especially if somebody has similar file types ready for use. In order for the MetaStock Downloader to be able to convert the CSV/XLS files to MS data type it will be necessary that the CSV files will have the proper header and data format: - Header Order = Date,Time,Open,High,Low,Close,Volume (the Volume field is filled with zeroes) Date format = 1/1/2005 Time Format = 13:15:00 (24 hours format with seconds field all zeroes) Open,High,Low,Close = any value, preferably above or equal to unity. The Time and Date will be computed as a function of the stimulus data length. I believe any fixed value for an IntraDay time frame will do (for the Date, Time computation) I will appreciate any comment and replies. Guara
bradulrich  
#2 Posted : Sunday, November 6, 2005 11:41:29 PM(UTC)
bradulrich

Rank: Advanced Member

Groups: Registered, Registered Users
Joined: 6/5/2005(UTC)
Posts: 35

Guara, You can create a static period sine wave like this in MS: [code:1:0259c10fff] x := 36; y := Sin(Cum(1)*x);[/code:1:0259c10fff] This will give you a ten period sinewave. I'm not sure the exact relationship between the x value and the period length, but you can play around with it and get an idea. I think this will be as good as the csv file. As for the other signals, I think the heterodyned signal is just a matter of adding two sinewaves ??? The other two are possible in MS I'm pretty sure. I can take a crack at them. We actually did not test the functions on those theoretical inputs. However, we have tested them on several theoretical inputs and alternative implementations of Ehlers' original code. Finally, I really like the ideas you have with the indicators. I have done some work is this area, but do not have any quantative results yet. In general, we (Mike may have some more to say on this) can speak more to the basic relationship of the functions to price. I think that Version 3.0 manual that we'll have any day now should help in this regard. One more point, I think we are going to release some versions of the functions that use different values as inputs, that is, when possible, we would allow a period value as input, and then we would perform the transformation internally. Much the same way an EMA allows a period input, but really uses an alpha factor, which is also an input to some of the functions. The reason we did not do this to start is that we wanted an "interface" that was exactly like Ehlers' functions, so as not to cause confusion. However, we notice that both implementations is the way to go. I'll let you know when those versions are available. Good luck, Brad Ulrich The Dynamic Market Lab, LLC www.thedml.com
bradulrich  
#3 Posted : Sunday, November 6, 2005 11:54:50 PM(UTC)
bradulrich

Rank: Advanced Member

Groups: Registered, Registered Users
Joined: 6/5/2005(UTC)
Posts: 35

the sinewave above has an amplitude of 1. To adjust the amplitude, a, try this: [code:1:07867d80b5]x := 20; a := 2; y := a*Sin(Cum(1)*x);[/code:1:07867d80b5] And try this for a heterodyned signal: [code:1:07867d80b5]x1 := 20; x2 := 12; inpt := Sin(Cum(1)*x1); inpt2 := Sin(Cum(1)*x2); heterodyned := inpt+inpt2; heterodyned;[/code:1:07867d80b5] To get a delayed signal, you can use the Ref() function [code:1:07867d80b5]refh := Ref(signal, -5); refh;[/code:1:07867d80b5] - Brad
Jose  
#4 Posted : Monday, November 7, 2005 1:54:25 AM(UTC)
Jose

Rank: Advanced Member

Groups: Registered, Registered Users
Joined: 1/19/2005(UTC)
Posts: 1,065
Location: Koh Pha-Ngan, Earth

Was thanked: 2 time(s) in 2 post(s)
bradulrich wrote:
x := 36; y := Sin(Cum(1)*x); This will give you a ten period sinewave. I'm not sure the exact relationship between the x value and the period length
Period length = 360 deg / x jose '-)
guara_riua  
#5 Posted : Monday, November 7, 2005 3:56:48 AM(UTC)
guara_riua

Rank: Advanced Member

Groups: Registered, Registered Users
Joined: 4/24/2005(UTC)
Posts: 77
Location: Canada

Brad/Jose Thank you for your replies. In fact I've got further with this issue: I've generated myself 3 diferent kinds of stimulus by using the "Free Pascal" compiler. Here are 3 files that implement 1,4 and 8 full cycles for the same frequency for each type of stimulus. Stimulus1 This is a simple chirped sine wave as described in Dr. Ehlers books. Stimulus2 This is a weighted 2 sine waves sum. Is a more comples stimulus and unveils some disturbing facts about how the PP+ external function behave. Stimulus3 This is a weighted sum of 3 sine waves that approaches substantially to the real price variations. It completes very nicely the results with PP+ mentioned above. I am attaching the zipped CSV files. For those that are new to the use of the format translation using the Downloader: - open each file with Excel and save it in the Excel format (XLS) - create a new equity in the Downloader with the same name (recommended) - convert the XLS file to MetaStock format in the folder of your choice (Should allow "Tranverse Directories" for sources and destinations and should check the "Create New File" check box. The period is 1 min, but it doesn't mater really) Brad, I believe these files could have a good use for any of your testings, in the case you don't have already something similar. I will attach a document that is embedding multiple samples of these stimulus files with PP+ indicators by replying to an older post of mine, relaterd to PP+. (Late Entry = Seems the server doesn't allow me to upload a +5MB file, limiting me to 2MB, even if the stated allowed file size is 8MB. I'll try to get some help from the "webmaster". Meantime I will attach the document i've wrote about the failures of some PP+ external functions) If anybody is interested about the code, just ask for it. I will be attraching it to a reply, if required. Just for those interested: i am not a professional programmer. I have used Turbo Pascal and similar compiler during my designer years to build up stimulus files for the hardware I was building. So don't expect the code to be very efficient or to look very professional. It is simple and can be ported easily to other languages. Guara
guara_riua  
#6 Posted : Monday, November 7, 2005 5:28:26 AM(UTC)
guara_riua

Rank: Advanced Member

Groups: Registered, Registered Users
Joined: 4/24/2005(UTC)
Posts: 77
Location: Canada

I have sent already all the stimulus files to Patrick, but he will be out for the whole next week. I believe anybody can get these after he will be back and we'll let us know were the files can be downloaded from. Guara
bradulrich  
#7 Posted : Monday, November 7, 2005 5:45:35 AM(UTC)
bradulrich

Rank: Advanced Member

Groups: Registered, Registered Users
Joined: 6/5/2005(UTC)
Posts: 35

I also see where you answered the sinewave period question the first time. Sorry i missed that. Thanks, Brad
Users browsing this topic
Guest (Hidden)
Forum Jump  
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.