[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[amibroker] Re: Optimizing and Robustness



PureBytes Links

Trading Reference Links

Ken,

These are your optimization, not mine, but no matter.

The ideas behind validity of system testing & optimzation, robustness 
etc are at best difficult.

A short reply ...

Some have stated that they are of the belief that writing, testing & 
optimizing a system based on some time period from 3/2000 forward is 
valid, but in the next breath also state that via some other means 
then the systems they are using they will know when the bear is dead 
and the bull has been reincarnated.  IMHO any of the bear market 
ralleys of the past 3 years could have in fact become full fledged 
bull markets and even if they some how couldn't have, some of those 
rallies amounted to 30% or more.  That's a hell of a DD to have to 
tolerate all the while somehow "knowing" that the bear is still alive 
and kicking.

--- In amibroker@xxxxxxxxxxxxxxx, "Ken Close" <closeks@xxxx> wrote:
> The attached 6 gifs show two aspects of the system Fred published
> several days ago.  I suggest you view them in order 1-> 6.
> 
> A1 and A2 are optimized from 9/1/1998 to 9/21/2001 and the equity 
curve
> in A2 shows the system "blowing up" in the OOS period from 
9/22/2001 to
> present.  No question that the optimized variables for this setup 
are
> not robust.
> 
> A3 and A4 paint a slightly different picture but perhaps not a 
different
> conclusion.
> 
> A3 and A4 are optimized from 3/1/2000 to 9/21/2001 and A4 shows 
that the
> equity line continues upward although with some higher variability.
> 
> A5 and A6 are more detailed views of the oos period and are labeled.
> 
> This raises the question that has been debated about including bull 
and
> bear markets in your optimization period vs only optimizing since 
the
> turn in the markets. 
> 
> It seems to suggest that the optimization period from 3/1/2000 to
> 9/21/2001 produces a more "robust" system---but I question whether 
this
> is so.  I am not sure I have confidence about when the performance 
of
> this system would deteriorate or "die" as has been stated.
> 
> My thanks to Fred for sharing his system and I hope this comparison
> stirs some continued good debate on this subject, from which we can 
all
> learn.
> 
> Ken


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/