The Rational Reminder Podcast · Thursday, October 1, 2026
Victor Haghani shared findings from an experiment testing if advanced knowledge of news could predict market returns. While macro traders performed reasonably well with a 60% hit ratio, even sophisticated AIs struggled with proper bet sizing, initially taking on too much risk. Haghani noted that while AIs improved after being trained on financial principles, sizing remained a challenge.
“And it's really hard to generate a good return. Turns out that on average, people didn't do very well despite having this newspaper a full day ahead of time.”
“And they had a good hit ratio, but they took too much risk. They didn't size things correctly.”
“But they did a really good job on figuring out whether the markets would go up or down and how strong. I mean, they were smart. It was impressive. But they got the sizing wrong to begin with, which was interesting.”