|
PureBytes Links
Trading Reference Links
|
Dave:
It depends on what you mean by "adjacent" (sorry for sounding like Bill
Clinton). If your parameter values in 3-D space signify a local maximum
(like, for example, ranging from 17 to 20 giving you profits of 2000 to
2200% where parameter values of > 21 or < 16 give you profits of 1000% or
less), then those nicely performing 'adjacent' parameter values won't be
robust, and you will likely suffer from losses when you use them in future
trading. However, if the parameter values give you a reasonably good
profitability from, say, 12 to 24 (like 1500 to 2000% or so), then you have
a fairly robust system that should perform well at any parameter value
between 12 and 24 without having to constantly re-optimize. The ideal is to
have a flat response area of the dependent variable against the parameter
values of the independent variables. And of course, as Fred said,
optimizing over both bullish AND bearish AND sideways periods is ideally
the best way to do it.
Do yourself a favor and pick up Robert Pardo's book on developing and
optimizing a trading system. He spends lots of time and space discussing
backtesting and walk-forward testing. I think it is an excellent resource
in this interesting subject area.
Al Venosa
Original Message:
-----------------
From: Dave Merrill dmerrill@xxxxxxx
Date: Thu, 30 Oct 2003 08:33:55 -0500
To: amibroker@xxxxxxxxxxxxxxx
Subject: RE: [amibroker] Re: Good & Bad Re-Optimizations (for Dave)
herman's ideas, while very valuable, weren't mechanical enough for this area
of my thinking. I wonder if anything mechanical like this can be truly
useful.
as I mentioned in my post to b, I also wonder why focusing on
nice-performing adjecent parameter values helps response to the reality of
the future, which is unpredictably changing data, not unpredictably changing
parameters.
dave
Presactly ...
One of the problems !? for most is that checking for robustness takes
longer. Eons ago Herman posted some fairly decent tools for this
which are along the lines of my own likings i.e. surface area plots.
This is pretty simple to do with systems that have two variables that
are being optimized on and requires more effort for more parameters
but is still doable. You can look at CAR or MDD or MAR or whatever
you like this way and get a good picture of where the cliffs are.
This shows pretty quickly where the most robust i.e. most likely to
succeed points ARE as opposed to the where the most profitable points
WERE. It takes longer then just taking a gander at a list of numbers
but imho it's worth the trip.
I liken the usage of where the most profitable points WERE as chasing
a constantly moving train that you'll never catch up to.
--------------------------------------------------------------------
mail2web - Check your email from the web at
http://mail2web.com/ .
------------------------ Yahoo! Groups Sponsor ---------------------~-->
Rent DVDs from home.
Over 14,500 titles. Free Shipping
& No Late Fees. Try Netflix for FREE!
http://us.click.yahoo.com/ybSovB/hP.FAA/3jkFAA/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/
|