[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [RT] Zigzag Statistics

PureBytes Links

Trading Reference Links
Thanks Sergey, Would love to see a future projection.
----- Original Message -----
From: Sergey
Sent: Thursday, April 17, 2008 6:23 AM
Subject: Re: [RT] Zigzag Statistics

Hi, Clyde
 
What you are describing is a childs approach to what the SwingMachine does in analyzing swings over whatever period using whatever data you want.
 
In this particular case I do not see any contradiction with SM, we do different things.
 
I simply provide the statistical analysis for zigzag for any time frame/data/zigzag parameters. As a result, I get diagrams like this one:
 
 
 
It shows the most probable resistance levels (red stripes in the right part).
 
My main interest is to get a reliable forecast, and we do that applying another technology - Back Testing. It is in brief:
 
a) break available price history on two sets training interval and testing interval ("out of sample")
b) provide spectrum analysis to reveal dominant cycles
d) generate neural network projection line based on these cycles
e) estimate how this projection line forecasts on "out of sample" price history data
f) conduct the same procedure for another "out of sample" data
 
We get a lot of projection lines (hundreds) for different models (with varying parameters in spectrum and neural nets), and we have a report for each one of them, like this one:
 
 
The technology is described here:
 
I think that for swing neural network analysis it is better to use detrended zigzag as output: http://www.timingsolution.com/TS/Uphistory/u_5.htm
 
As you see, Clyde, I am not discussing what Swing Machine does (I do not know that). I have enough things to do.
 
Best regards.
Sergey.
 
----- Original Message -----
From: clydelee
Sent: Wednesday, April 16, 2008 11:05 PM
Subject: Re: [RT] Zigzag Statistics

What you are describing is a childs approach to what the
SwingMachine does in analyzing swings over whatever
period using whatever data you want.
 
I commend you for your effort and particularly when
you say Fibonacci is BS.
 
If you want this published then email the SM list.
 
Clyde
----- Original Message -----
From: Sergey
Sent: Wednesday, April 16, 2008 6:08 PM
Subject: [RT] Zigzag Statistics

Hello,
 
I have done some statistical research for support/resistance levels for intraday data.
 
The analysed data are S&P500 futures, 1min bar from 1997 year. This is more than 2 millions price bars!
 
I analysed 0.25% zigzag, this is 1-5 hours swing (7000 turning points) and performed the statistical research for the ratio between down trend swing (BC) and uptrend swing (AB):
 
 
 
 
This is a histogram I have got for [down swing]/[up swing] ratio:
 
 
 
We can make following conclusions regarding the ratio of heights of down and up swings:
 
 
1) The most probable ratio is 1, 2, 3 i.e. whole numbers. In other words, the down swing is equal to up swing, or is twice higher.
 
2) Classical Fibonaci ratios do not work. They are marked by purple vertical stripes
 
3) Some clusters present at 0.87 0.93
 
4) Use of these criteria only to reveal next turning points is too risky
 
I did the same analysis for more complicated swing schemes like the ratio for down swing to previous down swing ( (B-C)/(A1-A) ratio).
 
Best regards.
Sergey.
 

__._,_.___

Your email settings: Individual Email|Traditional
Change settings via the Web (Yahoo! ID required)
Change settings via email: Switch delivery to Daily Digest | Switch to Fully Featured
Visit Your Group | Yahoo! Groups Terms of Use | Unsubscribe

__,_._,___