Responses and Info
Asia performs extremely well overnight: Hong Kong +5.8%, Singapore
+7.5%, Japan +2.5%, Korea +2.5%, Indonesia +5.7%.... European markets trading about 1% higher
across the board... S&P futures continue to
edge higher, with current level indicating a 45 pt gain on the Dow at the open.
SPLH beat estimates but predicted slower growth. It was at 23, expect that forcast to dump to below 15.
Dale Sinor - I didn't bother to do a rigorous estimate of profitability of the watch lists, so yes, its just shown from the first day of the post to the current day. I think the pointI was trying to make was that it was posted early into the climb, and remained on the long/bullish list for several days - long enough for someone to enter even if they only traded end of day.
I guess I'm not sure what was confusing about the post - I probably put to much info in there. Basically 77% of the watch lists improved an average of 5+%
As for how one uses news in a neural net, yes, it has to be quantified, but no, it doesn't have to be converted to a mathematical abstraction. I don't use a commercially available net, so don't compare that with something like AIQ or NeuroTrader (latter is excellent) or whatever. I created an interpretive text filter that works similar to a grammar checker (crude resemblance I would add). It essentially weights the news feed based on an assessment of positive or negative content. The tricky part though is coorelating that answer (say in the case of SPLH above with positive earnings growth and slower future growth) with possible market movement. How it actually works is proprietary....which also helps deflect criticism when its wrong (lol).
I suspect Neural Nets are boring reads for the thread, so it might be better to email me.
lastshadow |
