I have been trying to creat a stock prediction model for the past 6 years and I haven’t been able to create anything that does better than chance for stocks.
Two things have changed:
1. I predict only ETFs (lower variance)
2. New techniques have emerged relatively recently for neural nets.

The Value Metric is based on one week ahead prediction. When backtesting on last 5 years, the protections are ~95% within 1% of the true values.
The algorithm seem to be doing a good job finding local minima and maxima (hence buy/sell opportunities). Here is an example

Good tip. Didn’t know that

Welcome to blue sky. Seems to be a lot more civil here

I just noticed autocorrect did the same thing to my post

It’s from HeyGen app.

How do you mean

Still most of the AI and iOS community is on X.