|
PureBytes Links
Trading Reference Links
|
Gary, oops, sorry, I didn't see your post until now. My conclusions
are based on a lot more than anecdotal observations but I'll also tell
you that there's much confusion and disagreement among statisticians
about df, especially with models and data. Put 10 Ph.D.s in 10
separate rooms with your system you'll get 10 different answers on the
# of parameters that should be counted, how and why; their relative
impacts; and how curve fitted the system is given enough data to
produce X trades. Put them in the same room and they'd bicker like
bratty children. My $.02 is that any parameter that causes a
restriction or adds control contributes to curve fitting, whether it
comes from RUTVOL or Mars. So you should do all you can to minimize
them, no matter how much data you have. I'd agree that testing over
thousands of bars would be better than 52 as in your example. But as
I said in another post, this is a very difficult issue and my
preferred COA is bypass instead of frontal assault (Army lingo, lol).
Just remember there can be hidden parameters, some are worse than
others, and keep it simple. William Eckhardt says a system shouldn't
have more than about 3 or 4 parameters. I wouldn't disagree because
it's worked for me. About the academic discussion, I'd much rather
see a steady stream of system ideas.
--- In amibroker@xxxxxxxxxxxxxxx, "Gary A. Serkhoshian"
<serkhoshian777@xxxx> wrote:
> Mark,
>
> What you say is true about the effect of adding more parameters, and
I assume you are pulling that idea from the concept of information
theory where the information content goes up the square root of the
number of sources (2 independent sources = 1.4x info content, 3
sources 1.7x info content, etc.). Being as quantitative as you are,
I'm sure you are not basing your conclusions on anecdotal
observations.
>
> How are degrees of freedom impacted when you are pulling from
unrelated data sources like RUTVOL (i.e. russell price action and
nasdaq volume). Also, why is degrees of freedom such a deal-breaker
for you assuming you have a large sample size. For example, if you
have a 200 day ma, and you are trying to test data in which only have
1 year of data, then you have a problem as really you can only test
over 52 days that aren't required to make the 200 day ma. However, if
we are testing over thousands of bars the impact of that 200 ma is
lessened, isn't it? Thanks for the clarification here.
>
> BTW, when I mention that the discussion gets too academic on this
board, what I'm really speaking to is the fact that there is less of a
focus to share and generate trading systems on this board compared to
the FT-Talk board.
>
> There are quite a few folks on the FT-Talk board who don't have your
pedigree, but churn out system ideas on a regular basis. Some good,
some not so good. However, they are ideas, and the synergies achieved
by sharing those ideas result in tradeable systems.
>
> I'm certain you'd laugh at what you'd describe as a lack of due
dilligence on their part. The point is, the systems are
straightforward, the analysis is basic (yet effective), and the folks
who trade the resulting systems laugh all the way to the bank.
>
> Again, it's not about making $1 million overnight. It's about
steady returns.
>
> Regards,
> Gary
>
>
>
> quanttrader714 <quanttrader714@xxxx> wrote:
> Sometimes we're not academic enough and step into traps that are
> avoidable. Or is it better to skip the details, LOL? Degrees of
> freedom are lost for each parameter in a system that is used to
> determine a signal. So it depends not only on the number of
> parameters in the system, but also on the number of signals.
Imagine
> a perfectly curve fit system consisting of 10 buy dates and 10 sell
> dates for 20 parameters producing 20 signals with no degrees of
> freedom. This is a simple example and reality is much tougher. It
> can be *very difficult* to correctly account for all system
parameters
> and not only that, some may have significantly greater impacts than
> others. I personally choose to avoid this whole mess with very
simple
> systems that produce a lot of signals.
>
> --- In amibroker@xxxxxxxxxxxxxxx, "Gary A. Serkhoshian"
> <serkhoshian777@xxxx> wrote:
> > Howard,
> >
> > As always, you ask very relevant questions. There are several FT
> folks on the board who could tell you the exact dates when the
signal
> was developed. Even better, go to Werner Gansz' (signal creator)
> website, and ask him directly. He's a good guy, and was very
> responsive and informative in the few e-mails we've exchanged.
> >
> > http://www.madriver.com/~wwgansz/
> >
> > Taking this a step further Howard, how about walking us through
your
> analysis method on RUTVOL. Sometimes we get too academic on this
> board. So, here you have a complete signal that many call robust.
> Let's see how your method would be applied here, and my sense is it
> would add a great deal of value to a lot of us not so experienced
not
> to mention all the good karma you'll build : )
> >
> > Thanks,
> > Gary
> >
> > Howard Bandy <howardbandy@xxxx> wrote:
> >
> > Wow!! RUTVOL has a Lot of parameters.
> >
> >
> >
> > It is very unusual for a system with so many parameters to be
> robust. Does anyone know when the system development stops (the
last
> in sample date) and when the out of sample period begins?
> >
> >
> >
> > Howard
> >
> >
> >
> > -----Original Message-----
> > From: Gary A. Serkhoshian [mailto:serkhoshian777@x...]
> > Sent: Tuesday, November 18, 2003 1:07 PM
> > To: sgearhart12000; amibroker@xxxxxxxxxxxxxxx
> > Subject: [amibroker] RUVOL and BB system posted to AmiBroker
library
> >
> >
> >
> > Hi all,
> >
> >
> >
> >
> >
> > For easier access it's at this link in the AmiBroker library:
> >
> >
> >
> >
> >
> > http://www.amibroker.com/library/detail.php?id=307
> >
> >
> >
> >
> >
> > Kind Regards,
> >
> >
> > Gary
> >
> > sgearhart12000 <sgearhart1@xxxx> wrote:
> >
> >
> > Gary
> > Could you send me the 1 RUTVOL with BB Scoring.afl. I'm a FT user
> and
> > trying to use AB along with FT.
> > Thanks
> > Scott Gearhart
> >
> > ---------------------------------
> >
> >
> > Do you Yahoo!?
> > Protect your identity with Yahoo! Mail AddressGuard
> >
> >
> > Send BUG REPORTS to bugs@xxxx
> > Send SUGGESTIONS to suggest@xxxx
> > -----------------------------------------
> > 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.
> >
> >
> >
> >
> >
> > Yahoo! Groups Sponsor
> > Send BUG REPORTS to bugs@xxxx
> > Send SUGGESTIONS to suggest@xxxx
> > -----------------------------------------
> > 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.
> >
> >
> > ---------------------------------
> > Do you Yahoo!?
> > Protect your identity with Yahoo! Mail AddressGuard
>
>
> Yahoo! Groups SponsorADVERTISEMENT
>
> Send BUG REPORTS to bugs@xxxx
> Send SUGGESTIONS to suggest@xxxx
> -----------------------------------------
> 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.
>
>
> ---------------------------------
> Do you Yahoo!?
> Protect your identity with Yahoo! Mail AddressGuard
------------------------ Yahoo! Groups Sponsor ---------------------~-->
Buy Ink Cartridges or Refill Kits for your HP, Epson, Canon or Lexmark
Printer at MyInks.com. Free s/h on orders $50 or more to the US & Canada.
http://www.c1tracking.com/l.asp?cid=5511
http://us.click.yahoo.com/mOAaAA/3exGAA/qnsNAA/GHeqlB/TM
---------------------------------------------------------------------~->
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 http://docs.yahoo.com/info/terms/
|