SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Strategies & Market Trends : A.I.M Users Group Bulletin Board

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Jack Park who wrote (5255)8/2/1998 5:09:00 PM
From: JZGalt  Read Replies (1) of 18928
 
Hope that clears up any confusion.

I wasn't really confused so much as curious. What I am always concerned about when someone uses either an EP method or neural net or an optimized method is how well it performs after it stops learning and has to deal with real ups and downs in the market.

To clarify this thinking. Suppose I could fiddle, fool and optimize the various initial AIM parameters (cash percentage, buy SAFE and sell SAFE) until the cows come home and achieve the absolute maximum value you could have achieved over that time period using the AIM method. That doesn't tell me a thing about how it will perform in the future. What it does do is to give you some confidence that if the situation that existed in the time over which the AIM parameters were optimized continues to exists, then the "optimum solution" exists near the initial conditions that you started with.

What I've tried to do in the past is to see how the "optimized variables" change as more data becomes available.

The portfolios we put together for the oil service sector used approximately 18 months worth of data which is too short for any real results. I need to go back and plug in about 4 years worth of data and then see how AIM does on the last year if it was EP'd, trained or optimized on the first 3 years worth of data. <grin>

----
Dave
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext