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 : Market Gems:Stocks w/Strong Earnings and High Tech. Rank -- Ignore unavailable to you. Want to Upgrade?

To: LastShadow who wrote (10392)	6/19/1998 8:05:00 AM
From: LastShadow	Read Replies (1) \| Respond to of 120523

Someone messaged me some valid questions, I thought since I was gettting a lot of email to share it here - sorry for the length: "Do the profit percentages indicated anticipate optimal returns, entering long at the exact bottom of the move and reversing to short at the top? Also, are B/A spreads factored into your calculations?" No, the % returns posted do not reflect optimal returns. They reflect what was obtained (annualized) by using my trading strategy calculated based on entering either or long or short position at open, at market, with a $0.25 slippage, on the day following the signal from the net. Daytrading would (and did in the case of the stocks I live traded) provide better entries about 80% of the time. The slippage accounts for the bid/ask spread. "Do you generally recommend long and short entry points on stock lists provided by those who would like their stocks run through your nets?" I don't generally look at other's picks, but about 3 weeks ago I asked for some picks to test on the net in order to get a wider spectrum of stocks through it. I knew that I generally followed the same kinds of stocks, and wanted something I had no 'bias' toward to test. "Do you run your own watch lists through the nets as well? In your experience are there groups or types of stocks which are better suited to the nets than others?" I don't run my watch lists through the nets for the SI post - I do it the other way around. If a stock surfaces from the net, I check the TA. Actually I do look at the Fundamental, Technical and neural assessments of many stocks and generally go for ones with the strongest correlation from any two of them. "You had indicated in a recent post that the nets need to be tuned for specific equities. Are you suggesting fine tuning the nets for each stock will result in improved profits?" As for tuning the nets - this is a little more complicated to explain. A neural net is essentially a pattern recognition tool - one that, with other mathematical tools can be used as a predictive model. What you can attempt to predict is virtually unlimited - the close, the low, the open, the Exponential Moving average 8 days out, whatever you want. But in order to do that, one first has to identify what attributes are leading indicators of price. Is it the close and the high, or the volume and the stochastic, or the day of the week and the wilder oscillator? Many stocks have the same 'influencers' if we can call them that, just as some technical indicators work well for some stocks, and poorly for others. So what I do is first determine what are the primary elements that influence future price. Then I predict whatever it is I am looking to find out - usually close or high within the next few days, and load the historical data up to some point in time and have the net learn the patterns and train itself to predict with a minimal error over time. Then I establish a trading strategy that says buy when this event(s) occur and shell/short when something happens. I run the net to see what kind of return I get, and if I am not satisfied with the result I redo everything until I am. Once I am happy that I am geeting the best return I can, I then introduce virgin data (that the net hasn't seen) and forward test. If all of that works, I begin live trading. So Yes, one can optimize a net for a specific stock, although what I do is to find stocks that fit into a particular subset of influencers and trading strategy - saves a whole lot of computing time! "In your experience are there groups or types of stocks which are better suited to the nets than others?" I suspect that some stocks are more predictable than others, although it is more likely that I am not determining the right attributtes that influence movement, or not developing the right trading strategy. My assessment is always that I have not found out what it is, never that it can't be. One last comment - I was interviewed by Online Trading yesterday for an article in one of their forthcoming issues (its a apaer, not virtual magazine to debut next month) on using neural nets in financial markets. That article talked about a wider range of things, so when it comes out it might be worth picking up for those nights when you just can't fall asleep...lol lastshadow