The number that almost fooled me
Last year I started asking AI about stocks and got confident answers I couldn't check. This year I built a process around it.
Here's the kind of thing it catches. A YouTube video called a stock dirt cheap, and a data screen showed it at about two times free cash flow, the kind of number that makes you reach for the buy button. I was ready to buy. My AI research team flagged that the share count was likely from before a merger, which would make every per-share number look too cheap. Until the company's own filing confirms it, that number is Unknown, not cheap.
Another time, two sources disagreed on a price. It wasn't a data bug: one was quoting the Toronto listing in Canadian dollars, the other the New York listing in US dollars.
So I made one rule: if a number has no source and no date, it doesn't exist.
Opinions aren't a process.