|
PureBytes Links
Trading Reference Links
|
<SPAN
class=738511914-18102003>not that this is an "answer", but I'd say that using
all past history makes a system better at capturing long behavior patterns, but
slower to recognize changes. for example, if a stock went up for the first 10
years, then turned down, it will be quite a while before a model that sees the
past year as only 10% of the relevant data will turn. a model that intentionally
"forgot" sooner would turn sooner. it would miss larger scale pattern
repetitions completely though, due to its narrower examination
window.
<SPAN
class=738511914-18102003>
<SPAN
class=738511914-18102003>so maybe the optimimum examination period relates to
your intended holding period and trading time frame.
<SPAN
class=738511914-18102003>
<SPAN
class=738511914-18102003>dave
<BLOCKQUOTE
>
<FONT
size=2>From: Fred [mailto:fctonetti@xxxxxxxxx]<SPAN
class=738511914-18102003><FONT face="Courier New"
color=#0000ff>
I still believe that
the system that can make sense out of a whole bunch of history at once has
a much better chance of producing the desired results going forward then
one that wants to look at it a piece at a time as if the pieces were in
some way totally unrelated and could be some how described by some fixed
lookback or in sample period of time.
Yahoo! Groups Sponsor
ADVERTISEMENT
Send BUG REPORTS to bugs@xxxxxxxxxxxxx
Send SUGGESTIONS to suggest@xxxxxxxxxxxxx
-----------------------------------------
Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx
(Web page: http://groups.yahoo.com/group/amiquote/messages/)
--------------------------------------------
Check group FAQ at: http://groups.yahoo.com/group/amibroker/files/groupfaq.html
Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
|