Researchers introduce FinPersona-Bench, a simulation benchmark designed to measure Mandate Salience Decay (MSD), the phenomenon where behavioral mandates in autonomous financial agents lose influence over long deployment horizons. The benchmark uses a synthetic market that decouples price from fundamental value to evaluate three failure modes: trading without signal, panic-selling during crashes, and ignoring value during bubbles.

  • Evaluations of 18 frontier and open-source LLMs show MSD compounds over time and varies by model.
  • In crash scenarios, the behavioral gap between static agents and those receiving periodic mandate re-grounding grows 4.4x from the first to the final quarter.
  • Mandate re-grounding helps conservative agents in low-signal markets but worsens behavior for aggressive agents in the same setting.

The findings suggest that reliable long-horizon deployment requires selective, mandate-aware re-grounding based on agent profile and market regime.