Re: GEN - Description of Pattern Analysis used by Swing Machine (no gif)

[Bob Hunt <RHunt.066@xxxxxxxxxxxxxxxx>]

Clyde,

I like your work.  I had done some similar things with Merrill's work, but
certainly not to the degree that you've taken it.  I didn't work with the
concept long enough to develop much faith in its use.  But perhaps you
have.  What has been your experience in using it as a trading tool. Is it
successful? Do you offer the software to others?

For those of you who have an interest in this concept, and do not have the
programming skills that Clyde has (I certainly do not) you might want to
take a look at a couple of TASC articles that give you a starting point in
developing Excel spreadsheets that do similar kinds of pattern recognition
(again, not nearly to the degree that Clyde has taken it).  Take a look at
"Using Filtered Waves For Trend Analysis" in the July '95 TASC and "Pattern
Recognition And T-Bond Futures" in the August '95 TASC. Both articles are
written by Scott Barrie.

This type of study requires a 3 step process. 1) define and develop a system
for pattern recognition. 2) create a statistical base of past pattern
activity. 3) analyze current patterns in light of past pattern activity and
make predictions based on that analysis. The aforementioned articles do step
#2 for you by supplying a statistical base for one market (bonds). If you
want to use the technique for other markets you're on your own to create the
statistical base (which really isn't all that hard, but very time
consuming).  The thing that I like about Clyde's method is that it creates
the statistical base from whatever price series you feed into it.  That's a
great time saver!

I'd like to hear more about your work, Clyde.

Bob Hunt