the IBM harmonic

For discussion of how to use the Cycles Research Institute' own Cycles Analysis & Time Series software.

the IBM harmonic

Postby Eva » 01 Oct 2010, 11:33

Hi Ray

I saw on your blog the articel on harmonic on IBM shares.
http://cyclesresearchinstitute.wordpres ... are-price/

I would very much want to learn to do this thing in CATS. I know how to get the values from the spectral anlyis. But after that?
So I am asking for when you may have time, if you could put out the string of comands on how to do this.

Eva
Eva
 
Posts: 1
Joined: 01 Oct 2010, 11:19

Re: the IBM harmonic

Postby RayTomes » 02 Oct 2010, 22:50

Hi Eva

The initial cycles analysis was done with CATS spectrum command to get the many cycles periods.

After that a different program was used, based on Kotov's commensurability analysis, e.g. see http://www.springerlink.com/content/q760m75j18745071/

I will some time write articles about Kotov's method and supply some code for doing it in the forum here. It is not possibole inside CATS. It is very useful for studying many things.

Regards
Ray
User avatar
RayTomes
 
Posts: 128
Joined: 02 Aug 2010, 23:24
Location: Auckland, New Zealand

Re: the IBM harmonic

Postby Steve » 26 Oct 2010, 22:26

Hi Ray,

Can you please address that which appears to me to be undefined in your process
indicated at http://cyclesresearchinstitute.wordpres ... are-price/

From which is the quote
" Using CATS, the logs of prices were first calculated and then a spectrum was determined. In order of importance, the most significant (judged by Bartel’s test) cycles periods in days were determined accurately and found to be: "

Accordingly can you please show a list of those Bartels Scores associated with the statement
" ... most significant (judged by Bartel’s test) cycles periods in days "
I am looking for an answer that looks something similar to
http://img840.imageshack.us/img840/7017 ... 192010.jpg
which I recently used to emit the projection =
http://img819.imageshack.us/img819/2199 ... 192010.jpg
where I Subjectively used only Cycle Lengths that had
associated Bartels Score less than 0.03

Further, Why did you Select to Cut-off the Periods ( to be looked at ) at 67 ?
There was a criteria, so I am asking what was that criteria.

Further, Since the Data Set is Daily Price Data, I ask
What did you replace the Non-Trading Days Price with ?
NOTING : that in your article Titled " Copper Price Cycles Report "
at http://www.cyclesresearchinstitute.org/ ... cles.shtml
the prices you used, for non-trading day prices, were Linearly Interpolated
from the two nearest trading day Prices available. Now maybe it is not Trading Day
BUT the quote from the article is
" There are some periods of fixed prices and some periods where there is
missing data. Missing data has been replaced by linear trends joining to the
nearest known prices.
"

Generally specifically Why did you treat the Adjusted Daily Close
the way that you did treat it, and please indicate all steps that
you executed with the CATS software that are not documented
in the article about the IBM price.

Lastly, would you yourself utilize ONLY your above indicated 67 Cycle Lengths
in conjunction with CATS instructional Video 4
to place a BET on IBM ?

If your answer is NO, Then what further Criteria In-Your-Opinion are required
for BET placement ?


Regards,

Steve
Steve
 
Posts: 16
Joined: 13 Oct 2010, 21:25

Re: the IBM harmonic

Postby RayTomes » 27 Oct 2010, 02:58

Steve wrote:Hi Ray,

Hi Steve, sorry for slow replies.
Can you please address that which appears to me to be undefined in your process
indicated at http://cyclesresearchinstitute.wordpres ... are-price/

From which is the quote
" Using CATS, the logs of prices were first calculated and then a spectrum was determined. In order of importance, the most significant (judged by Bartel’s test) cycles periods in days were determined accurately and found to be: "

Accordingly can you please show a list of those Bartels Scores associated with the statement
" ... most significant (judged by Bartel’s test) cycles periods in days "

I have not kept an exact record of what I have done. When I tried to reproduce the results, I got something different. Clearly I have ignored cycles less than 3 days long. Maybe I used daily change in log or something. All I can say is that after 67 cycles the Bartels are 0.00035 so still very significant.
I am looking for an answer that looks something similar to
http://img840.imageshack.us/img840/7017 ... 192010.jpg
which I recently used to emit the projection =
http://img819.imageshack.us/img819/2199 ... 192010.jpg
where I Subjectively used only Cycle Lengths that had
associated Bartels Score less than 0.03

Further, Why did you Select to Cut-off the Periods ( to be looked at ) at 67 ?
There was a criteria, so I am asking what was that criteria.

It was sorted by Bartels and cut off. It seems that I also ignored cycles less than 3 days (there were very many of these). The reason for this is that cycles periods very near to 2 can often be aliases of shorter cycles.
Further, Since the Data Set is Daily Price Data, I ask
What did you replace the Non-Trading Days Price with ?
NOTING : that in your article Titled " Copper Price Cycles Report "
at http://www.cyclesresearchinstitute.org/ ... cles.shtml
the prices you used, for non-trading day prices, were Linearly Interpolated
from the two nearest trading day Prices available.

Yes, the same, linear interpolation.
Now maybe it is not Trading Day
BUT the quote from the article is
" There are some periods of fixed prices and some periods where there is
missing data. Missing data has been replaced by linear trends joining to the
nearest known prices.
"

Generally specifically Why did you treat the Adjusted Daily Close
the way that you did treat it, and please indicate all steps that
you executed with the CATS software that are not documented
in the article about the IBM price.

Lastly, would you yourself utilize ONLY your above indicated 67 Cycle Lengths
in conjunction with CATS instructional Video 4
to place a BET on IBM ?

In this case I was not specifically trying to make a forecast. I was trying to find out if there are a series of harmonically related cycles as Hurst Theory says and whether they agree with Hurst. I am sorry that I have not (yet) managed to exactly reproduce the same result. I will try again when I have more time.
If your answer is NO, Then what further Criteria In-Your-Opinion are required
for BET placement ?

If the broad structure of harmonically found cycles periods are accepted as real, then projections will be able to be made using assumed cycles near those ranges and keeping track of the phase of them at all times. By this I mean that a 20 day cycle phase 30 years ago is not relevant, only its phase of recent time --- enough time to get that phase correct.

I have not yet explored this deeply enough to give an answer as to how to make the predictions. However it is possible to make macros in CATS to show the phase of a nominal 20 day cycle (for example) and to see how rapidly it wanders about. Then we would know how much recent data to use for estimating the current phase of that cycle. We might also find cycles in the phase wandering. Some of this is rather unexplored territory. If it were not, it is unlikely that there would be money to be made.
:-)

This study was done to compliment The Hurst Trader software which uses Hurst periods of 40, 20, 10, 5 days etc. to see if that really was what was going on.

If you try to reproduce the results that I have in that IBM harmonic article, you will also need a program to do the Kotov type calculation. Here is a page with an example for the planets distances from the Sun: http://ray.tomes.biz/160min.html
At some time I will probably make such a program available with CATS, but my present program is in a script language which depends on particular software (Corel) so it needs conversion.

Regards
Ray
User avatar
RayTomes
 
Posts: 128
Joined: 02 Aug 2010, 23:24
Location: Auckland, New Zealand


Return to CATS Software

Who is online

Users browsing this forum: No registered users and 1 guest

cron