Learning Lineage & Performance Provenance

Verifiable Agent History

Every Permalator agent has a publicly verifiable learning lineage:

  • Memory commitments

  • Mutation events

  • Performance summaries

  • Regime exposure history

This allows users, DAOs, and capital allocators to evaluate agents based on long-term behavioral consistency, not short-term ROI.

Reducing Survivorship Bias

By preserving full learning history, Permalator mitigates survivorship bias common in trading bots where only winners are shown.

Agents are judged by:

  • Drawdown resilience

  • Adaptation quality

  • Regime survivability

  • Risk discipline

This elevates trust and transparency.

Last updated