PureBytes Links
Trading Reference Links
|
Al and
Owen,
I haven't seen the article in
question, but I was reasonably sure that it was all about 25 days of intraday
data and 25 bars of data. I would think that it is more important to
consider how many trades were generated than how many bars or days.
It's not a science, but if there were less than 25 trades per parameter then the
statistical significance is negligible (IMO). I would prefer to see
more than 100 trades per parameter, but 25 would be a
minimum.
<BLOCKQUOTE
>
-----Original
Message-----From: Al Venosa
[mailto:advenosa@xxxxxxxxxxxx]Sent: Tuesday, June 17, 2003 8:35
PMTo: amibroker@xxxxxxxxxxxxxxxSubject: Re: [amibroker]
Re: Annoying programming problem: ideas please?
Owen,
I am a subscriber to Active Trader, and I'm wondering if the article that
you are referring to is The Velocity System by Dennis Myers? If so, then it's
true that he tested his Least Squares Velocity System over a period
ranging from Oct. 9 to Nov. 8, 2002, but he used 5-min bars over that time
period. His in-sample (IS) optimization period consisted of 1 month of 5-min
bars while his OOS period consisted of 2 weeks of 5-min bars subsequent to the
1 month IS period. Since there are 78 5-min bars in a trading day and
approximately 22 trading days in a month, there were 78*22 or 1,716 bars
over which he optimized his system and 78*11 or 858 OOS bars comprising his
walk-forward test. So, it appears that he had sufficient data to provide a
reliable test, certainly not 25 bars. Do I have the correct article in mind? I
can't find any other instance where an author tested only 25 bars IS and 25
bars OOS.
Al Venosa
<BLOCKQUOTE
>
----- Original Message -----
<DIV
>From:
Owen
Davies
To: <A title=amibroker@xxxxxxxxxxxxxxx
href="">amibroker@xxxxxxxxxxxxxxx
Sent: Tuesday, June 17, 2003 3:28
PM
Subject: Re: [amibroker] Re: Annoying
programming problem: ideas please?
Fred wrote:> I hope that's not what Active
Trader is doing but IMHO 25 bars of> data is hardly enough to
accurately test anything on.It's sure the way I read their
articles. Cherry-pick a good-looking monthor six weeks, "test" it
on the next similar period, and cut things off whileyou can still
pretend your system is useful. I agree that 25 bars isnowhere near
enough for a test, and I consider publishing this stuff to belittle
short of fraud. Still, the programming required to do a
worthwhiletest would be interesting in itself.Owen
DaviesSend 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.
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.
|