AlphaCrafter: Multi-Agent Framework for Quantitative Trading
AlphaCrafter is a full-stack multi-agent framework designed for cross-sectional quantitative trading. It addresses the non-stationary nature of financial markets by integrating factor discovery, regime-adaptive selection, and risk-constrained execution. Unlike existing approaches that treat these components in isolation, AlphaCrafter unifies them into a coherent, rationality-driven pipeline. The framework aims to overcome limitations of static factor mining and execution systems that rely on role-playing agents with behavioral noise. By automating the entire quantitative strategy pipeline, AlphaCrafter seeks to maintain profitability across changing market conditions. The paper is available on arXiv under identifier 2605.05580.
Key facts
- AlphaCrafter is a full-stack multi-agent framework for cross-sectional quantitative trading.
- Financial markets are non-stationary due to macroeconomic regimes, microstructural frictions, and behavioral dynamics.
- Existing approaches optimize factor discovery, regime adaptation, and execution under static or isolated assumptions.
- Factor mining frameworks treat alpha discovery as a one-time search, assuming factor efficacy persists across regimes.
- Execution-oriented systems use role-playing agents that simulate trading committees, introducing behavioral noise.
- AlphaCrafter unifies factor discovery, regime-adaptive selection, and risk-constrained execution.
- The framework is fully automated and rationality-driven.
- The paper is published on arXiv with ID 2605.05580.
Entities
Institutions
- arXiv