MarthaStewart said:
Thanks for reminding me:
ALT
HITK
IGT
IMH
KBH
FCN
GVA
(GVA is probably more questionable on this one since I usually would only show ones that are over an avg volume of 250K)
For those that weren't paying attention, .4% from open to high of the day.
We test this via:
(HighPrice-OpenPrice)/OpenPrice = PercentReturn (well, technically we multiply it by 100 to get a percent)
ALT
0.057142
5.71%
yes
HITK
0.037091
3.71%
yes
IGT
0.005029
0.50%
yes
IMH
nope, down all day after open
KBH
0.018358
1.84%
yes
FCN
0.030493
3.05%
yes
GVA
0.022598
2.26%
yes
So we had 7 stocks, and 6 of them went up enough, 5 of them well over.
6/7 = 86%
And I think it was Matt that was pointing out that if a strategy is being used on a stock, then traders will pick up on the strategy and exploit it.
This is very true, and an excellent point since this is exactly why this works so well. This system doesn't just run once and then give you the algorithm - it retrains itself N times over a given period of time (they all vary - I'm playing with the adjustments on how frequently it needs to be trained - or rather updated).
So basically just a trader, well a good one, will pick up on certain strategies and exploit them, this is doing the exact same thing - and it changes at a rate faster than which anyone can pick up on the strategy.
Also - I keep saying this, but these that I have been demoing are not the best of all of the code - these are just the ones that I am using to show that the low end ones work pretty well and these are easy to show working.
The better ones do more testing than these and therefore make fewer predicitons, but are more likely to be right.
I won't be at work on Thursday or Friday of this week, and I'm not sure that I will be online - esp early enough to post these things.
I'm too lazy to get it setup to automate and post that way... I think - if I can get that working, then maybe they will be up - otherwise, I will be here tomorrow and then off (in terms of stock stuff) until Monday.