hey,
  can u pls provide me folowing:
  powerscan complete
  a newere update of beta version of amibroker?
  Thanks,
  Pass

Fred <[EMAIL PROTECTED]> wrote:
          Warren in his initial article on MEM suggests establishing as the 
beginning point a few bars prior to the beginning of the last change 
in direction ... This is fairly easy to do by determining swing 
hi/lo's of whatever magnitude one wants ...

--- In [email protected], "Paul Ho" <[EMAIL PROTECTED]> wrote:
>
> Mich,
> 
> I guess my last post really discuss two separate but somehow 
related issues
> 
> 1. How to programmatically determine the data length used for 
MEM
> 
> a. John Elhers discussed in his book Mesa and Trading Market 
cycles
> p.88 what he used to determine data length. In the beginning he 
maximised
> the SNR to establish to optimum data length. Later on he 
set "todays" data
> length to "yesterday" cycle length, assuming that it requires one 
cycle's
> worth of data to make a good measurement. Mich, you have said in 
your
> previous post that perhaps ½ cycle will suffice according to your
> experience. We of course can set today data length = n * 
yesterday's cycle
> period; There is also a practical problem in calculating dominant 
cycles. It
> is computing intensive for every bar. I wonder if it can be 
calculated more
> directly, for example using d(S(f))/df = 0. any views.
> 
> b. When compared the dominant cycle plot using MEM and that 
using
> Elher's formula , the result are quite different. MEM doesn't 
always
> produced a dominant cycle for every bar.
> 
> 
> 
> 2. How to deal with non-stationary signals
> 
> 
> 
> a) My knowledge about Hilbert is fairly limited (left 
University more
> than a quarter of a century ago) is mainly from reading Elhers' 
Rocket
> Science for traders. (pg 65) it is only about 10 pages – I am 
willing to
> photocopy it send it to if its relevant. Another site I've reading 
is
> http://www.hitech.technion.ac.il/~feldman/intro.html and
> 
http://lucid.igpp.ucla.edu/lessons/ess265/2005/hilbert_huang_transform
ation.
> pdf which provides some sort of introduction. EMD may be 
interesting. 
> 
> 3. The effect of downsampling on Spectral density
> 
> a) It seems that down sampling often provides interesting 
result and
> clearer spectral peaks which are otherwise masked. Have you tried 
it
> 
> 3. I'll have a look at your suggestions later on. Have you 
used any of
> it so far.
> 
> 
> 
> Cheers
> 
> Paul.
> 
> _____ 
> 
> From: [email protected] [mailto:amibroker-
[EMAIL PROTECTED] On
> Behalf Of Tom Tom
> Sent: Monday, 20 November 2006 4:04 AM
> To: [email protected]
> Subject: Re: [amibroker-ts] FW: Sample Data Range, Sampling Rate, 
and Window
> Effect on Au
> 
> 
> 
> Hi Paul,
> 
> What i like is the abiltity to see in real time the spectrum and 
see direct 
> how does it affect it to make detrending, moving average, etc... 
very 
> interesting.
> DSP plotting is at the end of the demo part and need to be de-
commented.
> 
> 2 times dominant cycle ? Niquist theory for sampling signal... ? 
(f_sampling
> 
> = 2*f_max_signal)
> Because without noise (and no more than 3-4 sinus added), good 
frequency of 
> cycle is catch with only HALF length (yes) of the data processed to 
built ar
> 
> model (based on experimentation i made on my work... nothing 
mathematical on
> 
> this héhé !).
> If we take 2 times dominant cycle, we can go back to FFT ; ) AR 
advantages 
> is that with very short data it can find cycles. But i agree with 
you in 
> noised signal it must be 2 times the frequency maybe.
> I got to experiment this too.
> 
> You are right, stationnarity can not be considered on long time.
> For exemple, speech processing (handphone GSM coding, speech 
> analysis/recognition, ...) is done with ARMA model 
(because "formant" - i 
> don't know the word in english, maybe the same ? - presents some 
peaks of 
> reqauncy in some colored noise).
> Speech is considered to be stationnary only a few milliseconds... 
so you 
> need to compute new coefficients all (20 millisec for exemple). But 
at the 
> end you finish with one second speech with 50 (1 sec/ 0.02sec) sets 
of 20 
> coefficients. It makes only 100 number to save... pretty damn good 
> compression !!!
> Trading data is considered to be stationnary... not at all. All 
difficulty 
> is here. Even there are considered not predictable if you don't 
trust there 
> is no cycle or pattern (i think personnaly there is sometimes cycle 
and 
> pattern and this make in our favor the 50/50 random).
> 
> I don't know a lot about Hilbert.
> Currently, in AR_Prediction.afl, i try to remove impulse from data 
(remove 
> end of day gap for exemple), but it is not the correct way so 
because one 
> very importante information for the intraday trading is removed... 
so maybe 
> yes, use thoses impulses (gap end of day, gap intraday, high volume 
peak, 
> high volatility peak, high spread bid/ask, many more sellers or 
buyers in 
> book order, ...) to construct new model.
> I thing an easy way based on the work already done on ar modeling / 
> downsampling / interpolation :
> 
> 1- take only quotes wich presents impulsion
> 2- discard all other quotes to empty
> 3- interpolate
> 4- predict AR
> 
> We can make the process for several stock (use composite for 
exemple) so we 
> got "market behaviour based on main events".
> 
> Should predict future data based only on main events/impulse ?
> 
> Can you explain me more about hilbert or give some source about 
hilbert 
> applyed in you idea (hilbert is very large concept).
> 
> Cheers,
> Mich.
> 
> ----- Original Message -----
> From: Paul Ho
> To: amibroker-ts@ <mailto:amibroker-ts%40yahoogroups.com> 
yahoogroups.com
> Sent: Sunday, November 19, 2006 12:58 PM
> Subject: RE: [amibroker-ts] FW: Sample Data Range, Sampling Rate, 
and Window
> 
> Effect on AutoRegressive Series was Polynomial Trendlines
> 
> Mich, Very handy work over there in http://www.amibroke
> <http://www.amibroke <http://www.amibroker.com/library/detail.php?
id=764>
> r.com/library/detail.php?id=764>
> r.com/library/detail.php?id=764
> 
> Just a thought or two on the sample length question. For stationary 
signals,
> the minimum sample length should be at least 2 times the dominant 
cycle. If
> we assume the minimum length is the optimum length. Then we can 
recursively
> set the sample length to 2 times dominant cycle + margin, until it
> converses. If it doesn't within a finite number of attempts, then 
signals is
> not stationary.
> 
> This brings in to another point that price signals are not always
> stationary. At times it leaves its stationarity when its excited by 
impulses
> which moves it another level until stationarity resumes. I am 
interested in
> Hilbert transform in providing instantaneous frequency during the
> non-stationary period. It might give us two mechanisms of 
prediction during
> different phases of the signal progression. Any thoughts?
> 
> /Paul.
> 
> __________________________________________________________
> Découvrez un moyen pour éliminez tous les virus de votre PC ! 
> http://safety. <http://safety.live.com/site/fr-FR> live.com/site/fr-
FR
> 
> 
> 
> 
> 
> [Non-text portions of this message have been removed]
>



         

 
---------------------------------
Check out the all-new Yahoo! Mail beta - Fire up a more powerful email and get 
things done faster.

[Non-text portions of this message have been removed]

Reply via email to