LLMs Fail to Beat Equal-Weight Baseline in Stock Trading

A recent study evaluating SOTA LLMs in stock trading reveals that no model outperformed a simple equal-weight baseline over a two-year simulation. However, models with stronger reasoning capabilities tended to incur smaller losses.

2026-08-01 ~ 2026-08-01 · 2 related posts

• 2-Year Test Shows SOTA LLMs Suck at Stock Trading — garrytan · 2026-08-01
• AI Stock Forecasts Underperform Even Split, ICML Paper Reveals — alejandroll10 · 2026-08-01