|
PureBytes Links
Trading Reference Links
|
This could be a religious issue.....no question.
Chuck LeBeau seems to favor ROBUST exit logic......which indicates to me it
is not trivial.
Your system approach sounds good, but that's what was done to attempt to get
DeMark's Sequential to be a decent system......the signals were OK, but had
enormous drawdowns.
Attempts to make the drawdowns "reasonable" resulted in a crappy system.
Maybe I didn't do it right or the signals were not the best ones for the
type of system.......
when is someone going to write a book on "How to Build Successful Automated
Trading Systems" ??
> -----Original Message-----
> From: Dennis Holverstott [mailto:dennis@xxxxxxxxxx]
> Sent: Tuesday, July 18, 2000 12:32 PM
> To: prosys@xxxxxxxxxxxxxxxx
> Cc: fritz@xxxxxxxx; Omega-List
> Subject: Re: A question about system design.
>
>
> Well, I agree, sorta. *IF* your system uses price-based calcs, they
> should be adaptive. However, I assume we're still talking about the
> sequential thingy. That's a pattern based system and there is nothing to
> optimize about a pattern. So, the FIRST step is to run it with no extra
> code whatsoever. Back to basics. Maybe just the pattern for the entry
> with a simple exit after x bars. The thing should be able to make more
> than it loses if you run it "naked". If you have to pile on a lot of
> entry conditions, multiple exits, etc. to make it profitable, chances
> are the pattern isn't any better than a random entry and you're just
> playing the overoptimization game. If, all by itself, it looks
> promising, then you can start adding all that extra stuff to fine tune
> it. Emphasis on the word FINE. The extras shouldn't become the system
> with the original idea pushed into the background.
>
> > Systems that either don't have dynamic adjustments for market
> conditions or
> > a single parm which sets other parameters when the market
> "changes".....are
> > asking for trouble.
>
> --
> Dennis
>
|