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: xu, b. who wrote (769)4/17/2000 1:12:00 PM
From: Optim  Read Replies (3) of 871
 
Bai,

I checked biocompsystems' site. Their live SP model is long. The recent out-of-sample equity curve is not great when the market nose dives. I am not sure how I should classify their model.

The model posted is merely an example. It has held up well for what it was designed to do. Biocomp doesn't actively research what underlying forces drive the market. They just build software.

All models, ANN based or otherwise eventually breakdown. The market dynamics are to blame. What works today fails tomorrow as methods are disseminated to the masses.

Perhaps a better example of what can be done with Profit can be found at Daniel Wilson's site, wildcattrader.com. He seems to catch the swings within a day or two.

I am wondering if Profit can allow me to easily scan 2000-3000 stocks and sort the result according to indicators.

Nope. For that matter I don't know of any ANN package that will do what you want. The computation times involved would be too long if you wanted to train individual models for each stock. If the net acts as a screener, you may want to consider just use NGO, the backend to Profit. It has a classification option which may work. I don't know how you would encode the screening criteria, but NGO does have a limit of 512 inputs. The base model with Profit is a 16 input version. Profit Pro ups this to 32 inputs.

I am interested in knowing if Profit can do stock screening. Or if TS Trader Pro can use equity curve as feedback to optimize models, then I will go with TS Trader Pro.

The only way I know how to do this with Trader is to use Net Profit as your genetic fitness function and then use a Trading Strategy with Ward's Adaptive Net Indicators as inputs. This would allow you to design crossover criteria, stops and filters, and then optimize the whole thing by Net Profit. The base product only allows you to optimize based on a simple % return calculation. It doesn't account for stops and such.

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