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 (830)9/14/2000 11:41:14 AM
From: mark_andrews  Read Replies (2) of 871
 
Hi Larry,

I have not found an easier method. I had to buy an Active-x net engine and start from scratch. The only genetic controls I know of are from BioComp and Wardsystems.

After studying how the genetic algorithm effects the stability of the model, I do not use it. I have found that neural networks in general lend no long-term accuracy due to the noise and instability they create. I hope everyone is familiar with how genetic algorithm works. Just in case there are a few who don't understand the internals, I will give a brief explanation.

A genetic algorithm treats each data sample as a gene. Instead of changing the parameters of the Technical Analysis applied to the model, it will change the data set by eliminating the defective genes. What is a defective gene? It is a data sample that does not fit the applied Technical Analysis. In turn, you do not know how many days of data are actually being used in the algorithm. Unless the technical analysis can be applied to all of the data samples and return appropriate returns for the effort, I don't see an advantage to using the applied technical analysis in the first place. If you force the issue by killing off the data samples that do not fit the TA, you are seasoning the model and it is destined for failure.

In summary, Neural networks in my mind are not any better than the applied TA. They have a knack for being able to filter out noise without lag, in return your models are short lived and very unstable.

Regards,

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

HomeMailHotSubjectMarksPeopleMarksKeepersSettings
Terms Of UseContact UsCopyright/IP PolicyPrivacy PolicyAbout UsFAQAdvertise on SI
© 2025 Knight Sac Media.  Data provided by Twelve Data, Alpha Vantage, and CityFALCON News