EvolveTrade: Self-Evolving LLM Trading Agents Rewrite Their Own Policies

A research framework that lets an LLM-based trading agent self-evolve its tool-use policy by reviewing its decision traces and portfolio outcomes, while keeping the backbone model fixed.

Standard agents rely on static, hand-written tool-use policies fixed before deployment, whereas EvolveTrade periodically rewrites that policy based on realized performance.

The results are early-stage and academic, but they point toward more adaptive trading and risk agents; teams should validate in simulation before any production use.

Sharpe Ratio and Cumulative Return, with improvements in most evaluated settings across multiple market regimes and two LLM backbones.