I am in the process of creating a new video for my YouTube series about using AI to make stock trend predictions. The next video will be about a algorithm I developed called WAEVE. While getting the algorithm ready for the video I ran a test and it produced a rather serious bearish chart for $SPY.
So I thought that can't be right and tested it with some real A.I. code and it produced the following charts:
First using the RandomTree classifier:
and another using the RandomForest classifier.
First using the RandomTree classifier:
It will be great to see in 30 days time how accurate you were relative to the above prediction. Something that all journalists and bloggers should do
ReplyDeleteI found an error in the software; the code was using the price data backwards; so it would use prices from say 30 days ago to compute a predicted price 30 days ahead, 29 days out for 29 days ahead, 28 days out for 28 days ahead,... etc. So that would force the expected price to be going down the more it went out into the future since SPY's price farther in the past is much lower than the current.
Delete