Paper: arXiv 2604.02279

Authors: Andrew Ang, Nazym Azimbayev, Andrey Kim

Abstract

Agentic AI shifts the investor’s role from analytical execution to oversight. We present an agentic strategic asset allocation pipeline in which approximately 50 specialized agents produce capital market assumptions, construct portfolios using over 20 competing methods, and critique and vote on each other’s output. A researcher agent proposes new portfolio construction methods not yet represented, and a meta-agent compares past forecasts against realized returns and rewrites agent code and prompts to improve future performance. The entire pipeline is governed by the Investment Policy Statement–the same document that guides human portfolio managers can now constrain and direct autonomous agents.

Complexity vs Empirical Score

  • Math Complexity: 2.5/10
  • Empirical Rigor: 5.0/10
  • Quadrant: Street Traders — practical and empirical, lighter on theory

Why this score: The paper introduces a highly novel multi-agent LLM framework for end-to-end portfolio management, including dynamic code-rewriting and self-reflection loops. While mathematically light—relying on standard portfolio construction routines via agent execution—the conceptual framing and practical application of LLM orchestration to institutional workflows are exceptionally innovative.

Research Flowchart

  flowchart TD
    A["Research Goal: Autonomous Agentic Strategic Asset Allocation Pipeline"] --> B["Input: Investment Policy Statement (IPS) & Capital Market Data"]
    
    subgraph Pipeline["Agentic Architecture (~50 Agents)"]
        B --> C["1. CMA Agents & 20+ Portfolio Builders Generate Strategies"]
        C --> D["2. Peer Critique, Voting & Researcher Agent (Proposes New Methods)"]
        D --> E["3. Meta-Agent Rewrites Code/Prompts Based on Realized Returns"]
    end
    
    Pipeline --> F["Outcomes: IPS-Compliant, Self-Improving Multi-Asset Portfolios"]