Silicon Investor (SI) -- The First Internet Community

STOCKTALK

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 : NeuroStock -- Ignore unavailable to you. Want to Upgrade?

To: Len Giammetta who wrote (619)	1/14/1999 6:01:00 PM
From: CatLady	Read Replies (1) \| Respond to of 805

>> " I had thought that all that stuff about the look backs and influence periods related strictly to the data in the training period. It's been my understanding that the data in the verify period in no way influences the prediction until the prices slide into the training period, and is treated as if it isn't even there until that time. " Just change that one word in the above quote, prediction to *training* to correctly describe the process. The current day's data is used to make the prediction, based on patterns learned in the training period. >> "Pretty stinky market today, huh?" So far, I'm pleased with the way my newer nets are handling this week's correction. Got my fingers crossed that they'll continue working.

To: Len Giammetta who wrote (619)	1/14/1999 8:16:00 PM
From: Jay Hartzok	Read Replies (3) \| Respond to of 805

Len & CatLady, If both of you are right, then either I am very confused or don't understand English very well. Excerpt from the help files: When you hit the PREDICT button the network automatically uses ALL the data it needs to predict the stock's behavior including the latest data, not the dates in the training dialog box. This is from the FAQs; response to the second question. I say once again, just a bit differently, that Neuro's current predictions are calculated from the influence of the previous day's data, {data used from whatever the influence settings are for each input}, based solely on the patterns that it has found in the training period. Think about it. If you have a 4 month verify period, how could Neuro ever possibly predict tomorrow's price movement based on data that is 4 months old. The price of the stock could have changed drastically by then. It has to use the last few weeks of current data to formulate the current prediction. It has to compare the most current data to the patterns it found in the training period to get the tomorrow's prediction. {Current data does not mean just today's numbers. It means all of the necessary data that is required to make a prediction. You have set these parameters with the influence settings} So, if you have an influence setting of 15 for s/t filter and your index data for the last one to three weeks is missing, then the tomorrows prediction is invalid. The prediction will not be valid again until the 15 days has elapsed and Neuro once again has all of the necessary current data it needs to formulate a true prediction. If I am in error on this, please enlighten me. Let's clear this up right away. Jay