Agentic Quantitative Trading: A Survey of Workflows, Systems, and Evaluation
This survey reviews agentic quantitative trading, covering five stages: factor mining, signal discovery, portfolio construction, execution, and risk management. It highlights that current systems focus on signal discovery, multi-agent reliance on aggregation, and that strong predictive capability does not ensure live trading performance.
01 ABSTRACT
This paper is a survey by researchers based on existing literature and systems. The authors argue that quantitative trading is moving from isolated predictive models to agentic workflows. Facts include current systems concentrated on signal discovery, multi-agent dependence on aggregation, and benchmark evidence showing a gap between model capability and live trading. Opinions include future needs for full workflow integration, stronger coordination, and matched evaluation.
02 KEY FINDINGS
- Agentic quantitative trading covers five stages: factor mining, signal discovery, portfolio construction, execution, and risk management.
- Current systems are concentrated on signal discovery; full workflow integration is uncommon.
- Multi-agent systems rely heavily on aggregation, lacking diverse workflow structures.
- Benchmark evidence shows strong model predictive ability does not reliably translate into live trading performance.
- Future directions include more complete trading workflows, stronger coordination, and matched evaluation.
AI GENERATED SUMMARY / DISCOVERED BY ARXIV Q-FIN