Jay and all: Just began reconstructing a net I had on AMAT with very interesting results. My original net had a few relateds, stocks that traded in tandem from the same industry group. I fed the net l/t and s/t filters and utilized all the settings on all the relateds. The training scattergraph was very sharp, the net fit. I had no verify.
I just remodeled it using 6 indices, from industry, sector, and general market, and used only l/t filters. So far I have about 2 hours on it and it is fit, and accurate, per the verify, and is training very quickly (converging nicely). I kinda expected all that from prior results using indices. What I didn't expect was that the net, only after 2 hours, is about 50% more profitable! The original indicated profits of about $70,000, the new one is already at $110,000 and still climbing. Just some food for thought.
By the way, I'm interested in adding another 15 stocks to my portfolio. I'm interested in volatile issues, in any industry, who's net exceeds $100,000 of profits when trained. I would appreciate recommendations so I can spend the next few weeks training them. I don't need the relateds, as I like to pick them myself, though, just the stock. Thanks in advance. |
