Page 1 of 1

Reoptimization( Model Shelf Life )

Posted: Tue Apr 25, 2006 12:42 pm
by freeman
Is it true to say the size of the test window has a affect on shelf life?

Systems that use optimization require some type of periodic reoptimization to fine tune the trading model to current market conditions.
Short test windows require more frequent reoptimization,(they have shorter shelf-life),and clearly longer test windows require less.

General Rule:
For the size of the trading window is between 1/8 and 1/4 of the test window.
So if a 24mth test window is used to optimize the system,then it can safely be traded forward for between 3 (24/8 = 3 ) and 6 (24/4 = 6 ) mths.

Any thoughts ?

Posted: Tue May 09, 2006 10:54 am
by Forum Mgmnt
I'm sure that there is something to this idea. Markets go through phases (smooth trending, consolidating, range bound, volatile but trending, etc.) and strategies that are dependent on a market being in a certain phase will not last as long as strategies that work independent of that phase.

I don't know how this might translate into the type of rule of thumb you list, however. I think the lifespan of a strategy has more to do with the way it makes money and the types of markets it requires to do so than with the timeframe of optimization.

- Forum Mgmnt

Posted: Mon Dec 18, 2006 3:18 pm
by freeman
Hi Forum Mgmnt,
sorry for the ultra slow reply(oops). I got this from Robert Prado book System Design and Optimization of trading Systems.
I have no idea how he gets hes formula for this,but i found it essential in short term systems.It has been a while since this post and i have learnt to use multiple sytems for the various conditions u mentioned.

thanks freeman