I can’t help thinking that if technical analysis indicators worked particularly well then machine learning algorithms would have cleaned up years ago. The fact they haven’t suggests it only works until it doesn’t. However I would think the more supporting information you can feed in the more likely it is to work eg scraping the news for sentiment, announcements, sector trends, etc, but that’s where large scale compute resources would be needed. Well, that’s my novice experience when playing around with it where something that worked on training/test datasets never panned out in application. So I tend to skip all that now and just invest for the long term… seems a lot less effort and, with the aid of some very good forum posts here and a bit of research, more lucrative.
4 Likes