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 : Neural Nets - A tool for the 90's
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Larry Livingston who wrote (664)9/27/1999 9:39:00 AM
From: Optim  Read Replies (2) of 871
 
Hi Larry,

I'm curious if that's the addin that you're using and if it is could you let us know how good the documentation and customer service is and how easy it is to implement. Can you basicaly fine tune the frequencies you want to look at or are you locked in to what they give you?

The addin for Excel is what I use mainly, although the NS Trader version works well (but sloooow because of Trader's indicator calculation times!) and can be genetically optimized. I haven't had to use the tech support yet as the product is pretty straight forward and the documentation is absolutely excellent. It spans over seventy pages and describes all options and includes sections on the theoretical benefits to use these forms of Wavelets.

The addins allow you to specify all periods used. I have found that isolating a few short-term and one long-term has given me the best results thus far. A neat thing to do is just let Trader optimize the indicators on a very long stretch of data, say five years or more. Then apply these figures out of sample on a hold back period. The results are often quite good and remain consistent for most issues.

I would also be curious to know if you are feeding your nets just the raw Wavelet signals or whether you are deriving indicators from them.

I have played around with momentums of the wavelets as well as a few other technical inidcators but the results are usually the same, so I'm not sure there is any benefit to it. Just the raw wavelets are often quite good, although I find I get much better results when the neural network structure takes into account lags of the indicators. For example BioComp's NGO supports TDNN and CATNN structures which automatically determine and add optimal lags to the network structure. The results are often quite good.

I just wish that the Morlet/Gabor Wavelets were integrated into Profit/NGO to make things easier. I've had to become somewhat of a Visual Basic guru to make things work without taking days to do. So much work, so little time... :)

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

HomeMailHotSubjectMarksPeopleMarksKeepersSettings
Terms Of UseContact UsCopyright/IP PolicyPrivacy PolicyAbout UsFAQAdvertise on SI
© 2026 Knight Sac Media.  Data provided by Twelve Data, Alpha Vantage, and CityFALCON News