|
PureBytes Links
Trading Reference Links
|
On Sat, 03 Jan 1998 13:41:13 -0800 "bobrabcd" <bobrabcd@xxxxxxxxxxxxxxx>
writes:
Ron, how about doing a correlation of OEX for 12-09/96 to 1-2/97 with the
12/09/97 to 1/02/98 so that it covers the same holiday period?
Bob, the correlation study covered OEX data from 12/02/96 to 1/02/98. I
used 3,5,8,13 and 21 day periods to find correlations. If all time
periods had a better than 50% correlation I considered that a high
probability match. Than I would monitor the correlation as the december
time frame unfolded. If the correlation continued to be high I would look
for similar comparisons with the indicators I have developed with the
correlated time periods. I have found this to be quick way of finding
cyclical phenomenon that might otherwise not be obvious. My idea was that
if cyclical is persistent in mkt data than all I would have to do would
be to find the last time the cycle occurred, check for a match with
current data, and see what happened last time the pattern made its
appearance. This method also eliminates any bias toward a seasonal
opinion on the mkt. ie: the match of mar. and aug. with the current dec.
mkt action. I will eventually expand the data base back to 1987. Maybe I
will be able to come up with a pattern recognition data base. Not to get
off the subject, but I can now read .gif files sent to my juno email. I
save the email as text, strip off the text down to the attachment, decode
with xfer and then read the file with the aol program (at last aol is
useful for something). It saves me the hassle of asking people to send
files to my aol address (which is impossible to log onto at a convenient
time).
Thanks
Ron McEwan
|