Subject: Re: OT: Using AI for Initial Stock Analysis

Gwen Hofmeyer (Maiden Financial) is starting to win a reputation for extraordinarily thorough, painstaking research. For example, in looking at Ingles Markets' property holdings she didn't stop at the balance sheets -- she went through county tax records. On a first-order analysis it sounds like the kind of human data-gathering that might be 'disrupted' by AI, but she's identified a strategy I think Charlie would approve of: use AI to identify the questions that AI can't answer, which then become candidates for research.

https://www.linkedin.com/in/gw...


How I use AI for research: To identify information that is not publicly available, and therefore not retrievable through AI's use. Case in point:

I asked ChatGPT, Perplexity, and Gemini the following question:

"How many US and Canadian publicly traded "industrial" categorized companies with market capitalizations of $400 million or more have made at least three acquisitions since 2017?"

An abridged summary of all three platform's responses:
"I cannot provide an exact number."

Task acquired: Manually analyze all 412 US and Canadian industrial-categorized companies with >$400MM market capitalizations in "yes/no" format, with all "no's" failing the 3+ acquisition test serving as a foundation for further research (indicative of unusual activity within a very acquisitive category).