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 -- Ignore unavailable to you. Want to Upgrade?


To: Optim who wrote (671)10/6/1999 8:06:00 PM
From: Optim  Read Replies (1) | Respond to of 871
 
Some thoughts on inputs...

Lately my focus has been on examining individual stocks for any seasonal bias. To explore this I created a spreadsheet in Excel. For each day of the month, for all months, I total the percentage gain (or loss) and then take the average of this amount for each day. I then graph this in Excel and look for any discernible patterns. What I have notice in a few issues (CSCO and a few semis) is a strong begining of the month rally, followed by a dip, then a run up to the day before options expiration, then a dip, then a mild rally until it tanks at the end of the month.

Now I know there are much better ways to calculate seasonals, and I just wanted to throw together something rough to see if it is worth investigating. From my first impressions it is, as a simple system that buys the Monday open before options expiration and sells the Thurday close would yield close to 10% per trade on a few semis that I looked at. I haven't coded it in Metastock or SuperCharts to get a full trade report yet, but it looks promising.

This could be one area to explore. Has anyone done similar work that they would care to share? Or does anyone have any recommendations of other forms of seasonals we could explore?

I have been thinking of how to code this as an input to a neural net so as to highlight the period. I have programmed a simple Excel VBA macro that counts the number of days left to expiration. I have more testing to do, but the genetic process in NGO hasn't selected them as a good input thus far. Any recommendations on how I could highlight this period? Now that I think of it, perhaps a simple binary input to highlight options expiration week would suffice. I'll have to test it out.

Daniel



To: Optim who wrote (671)10/6/1999 9:12:00 PM
From: Gary E  Read Replies (2) | Respond to of 871
 
I would be interested in a basic beginers discussion, starting with page 1 ...one that would work with the S&P futures...

I have read very little about nets that was understandable..so maybe I am confirming ignorance here but I think it may be because I dont know where to start, and dont want to part with big bucks for a wizzard data program that may or may not be worth 2 cents...or what I have read is so far overhead its like trying to shoot ducks at 400 yards,,,fugitaboutit....

So if you want to start on page one ,,,I'm all ears..

thanks

Hal



To: Optim who wrote (671)10/7/1999 11:48:00 AM
From: Eugene G. Glover  Read Replies (1) | Respond to of 871
 
Optim,

I have tracked (lurked) this thread for some time as I have a interest in developing systems of my own. I noticed that it is quiet much of the time but I have picked up some interesting tidbits. I have the neural net program Brain Maker Pro but have only done some of the tutorials and have not actually done any system building of my own yet. I just received WaveFin which I learned about here. Since I am in the formative stages of my own work I greatly appreciate your efforts to activate this thread, especially at this time. Thanks for your efforts.

GENE