Also, one more thing I’d like to point out is that, according to your argument earlier (that is, this works -> many millionaires). We must have 0% people with skin cancer because skin cancer now can be detected with more than 95% accuracy, and there are many startups/companies spun around that. Solving a machine learning problem doesn’t directly translate to its desired effect.

I agree with your second claim (we cannot predict the future movement based “purely” on historic data). But I disagree that it is impossible. On the other hand, you are making high level claims (e.g. we cannot predict correctly most of the time) without factual data. I can discuss further if you can bring in machine learning concepts (e.g. model overfitting/ bias / unclean data), pointing out to the result that indicating lack of prediction capability of the model.

Otherwise, I think the value this conversation providing to the reader is very limited. I’ll be happy to discuss this further personally, if you like, and feel free to contact me at thushv@gmail.com.