True story. IRL I have over 500 pages of written material publicly available on investment philosophy taken from reading pretty much everything written on finance since Pascal and de Fermat formulated probability theory. A fund manager took all of that and trained an AI to be a “Jim Kozlinski” clone and fed a FactSet data API into the model. Backtested results show a very clear quintile breakdown of most fitting to least fitting with most fitting outperforming by 400 bps over 30 years, and least fitting underperforming by 600 bps. AI can do the work, it’s just trained on garbage, and will continue to be garbage until the conventional wisdom changes. No this is not my real name.