|
PureBytes Links
Trading Reference Links
|
Clyde,
As I become more comfortable with your Swing Machine, I've started experimenting with continuous futures contracts. Most of
the contracts that I've looked have such short, frequent, volatile swings that I find it hard to place much faith in
forecasting based on the nature of those swings. They just seem too erratic to count on predictions based on that erratic
behavior.
Except for T-Bonds. I was first introduced to this type of pattern recognition as applied to T-Bonds several years ago. I
recognized the potential of accurate forecasts based on this idea, but never took it to a level that would be practical and
usable on a daily basis. I thought I'd take a look at T-Bonds again with the aid of your software and see what I came up
with.
A quick review of several different time periods indicates to me that there is some forecasting capabilities here, and that it
could indeed become an aid in trading this market.
However, I did get some interesting results when run on a continuous contract that goes back to 1978. Attached are two gifs.
"usdaily.gif" clearly shows a down forecast starting immediately and lasting about 5 days. On the other hand, a run of weekly
data, "usweekly.gif" appears to me to be depicting a market peak in the current time frame, but the lack of a forecast very
far into the future leaves me a bit confused.
What's your take on this and what do you think these pattern recognition results are telling us?
Bob Hunt
Attachment Converted: "c:\eudora\attach\USDAILY.gif"
Attachment Converted: "c:\eudora\attach\USWEEKLY.gif"
|