|
PureBytes Links
Trading Reference Links
|
Owen wrote: I do understand the difference between testing five-minute
bars for a monthand testing EOD for the same period, and no doubt I should
have included thebar length in my original message. Nonetheless, I am
profoundly skepticalthat two month's worth of data constitutes a valid
study, no matter how manybars it's broken into. This stuff stinks of
cherry-picking to me, and formy own satisfaction I'd like to be able to
prove it.
But Owen, I'm not defending Myers
in any way, but to truly compare apples with apples, if you consider just
the no. of bars and forget for a moment the time period over which those bars
occur, then the true comparision is between a month's worth of 5 minute bars
(1700+ bars) and 1700 days of EOD data (~ 6.5 years), not 22 days of EOD data.
In other words, testing your system over a month's worth of 5-min data is
equivalent to testing your system over 6.5 years worth of EOD data, isn't it
true? So, from a purely statistical standpoint, the no. of bars is the same, and
therefore the validity of the system results should be the same. Now, if you
want to consider the overall magnitude of a ticker's price change, then perhaps
your skepticism is justified, since the price change in a month is likely not
going to be in any way similar to the price changes that would occur in EOD data
over a 6.5 year period. <FONT
face="Times New Roman">
<FONT
face="Times New Roman">Aside
from which, who knows? Someday I might have a use for
walk-forwardtesting. It's an interesting technique, and I'd like to
know how to do itwithin AB.
I do walk-forward testing all the time
in AB simply by optimizing over a period of, say 1/2/98 to 12/31/01 (which
covers both bull and bear markets), then forward testing from 1/2/02 to present
time, using EOD data. I feel this is quite valid and seems to work well for
me.
Al
Venosa
Send BUG REPORTS to bugs@xxxxxxxxxxxxxSend SUGGESTIONS to
suggest@xxxxxxxxxxxxx-----------------------------------------Post
AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx (Web page: <A
href="">http://groups.yahoo.com/group/amiquote/messages/)--------------------------------------------Check
group FAQ at: <A
href="">http://groups.yahoo.com/group/amibroker/files/groupfaq.html
Your use of Yahoo! Groups is subject to the <A
href="">Yahoo! Terms of Service.
Yahoo! Groups Sponsor
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.
|