I've discovered a very interesting thing, which apparently most in the machine learning space don't realise (I didn't until recently). Trading isn't what most people think it is at all. I've spent a week or two doing a bit of reading into what it actually is.

Three things that have jumped out at me:

1) Trading is not investing. I was extremely doubtful about a lot of the "models" people throw around in this area because to me they seemed closer to numerology than anything related to investment. Then someone pointed out that trading is not investing, and suddenly I looked at these models in a new light.

2) There's lots of interesting trading you can do even without having any idea about what the price is going to do. Do you think volatility is going to increase? You can make money from that - read about the Collar Trade Strategy (This is just an example).

3) There's lots of strategies which don't make enough money for companies to be interested in, but are viable for an individual.