Design and Empirical Study of a Large Language Model-Based Multi-Agent Investment System for Chinese Public REITs

Paper: arXiv 2602.00082

Authors: Zheng Li

Abstract

This study addresses the low-volatility Chinese Public Real Estate Investment Trusts (REITs) market, proposing a large language model (LLM)-driven trading framework based on multi-agent collaboration. The system constructs four types of analytical agents-announcement, event, price momentum, and market-each conducting analysis from different dimensions; then the prediction agent integrates these multi-source signals to output directional probability distributions across multiple time horizons, then the decision agent generates discrete position adjustment signals based on the prediction results and risk control constraints, thereby forming a closed loop of analysis-prediction-decision-execution. This study further compares two prediction model pathways: for the prediction agent, directly calling the general-purpose large model DeepSeek-R1 versus using a specialized small model Qwen3-8B fine-tuned via supervised fine-tuning and reinforcement learning alignment. In the backtest from October 2024 to October 2025, both agent-based strategies significantly outperformed the buy-and-hold benchmark in terms of cumulative return, Sharpe ratio, and maximum drawdown. The results indicate that the multi-agent framework can effectively enhance the risk-adjusted return of REITs trading, and the fine-tuned small model performs close to or even better than the general-purpose large model in some scenarios.

Complexity vs Empirical Score

• Math Complexity: 3.5/10
• Empirical Rigor: 6.0/10 (capped from 7.5: transaction costs only partially modelled; see penalty flags)
• Quadrant: Street Traders — practical and empirical, lighter on theory

Why this score: This paper presents a complete, practical multi-agent LLM framework applied to the low-volatility Chinese REITs market, backed by an open-source implementation and solid backtesting. While mathematical complexity is limited to heuristic rules and basic statistics, the system design, empirical setup, and model comparisons provide strong practical value.

Research Flowchart

  flowchart TD
    A[Goal: Enhance Chinese REITs Trading via LLM Multi-Agent System] --> B[Input Data: Market Prices, REITs Announcements & Financial Events]
    B --> C[Step 1: Multi-Agent Analysis<br/>Announcement, Event, Momentum & Market Agents]
    C --> D[Step 2: Signal Integration & Directional Probability Prediction]
    D --> E{Prediction Pathways}
    E -->|Pathway A| F1[General LLM: DeepSeek-R1]
    E -->|Pathway B| F2[Specialized SLM: Fine-Tuned Qwen3-8B SFT+RL]
    F1 --> G[Step 3: Decision & Execution Agent<br/>Generates Position Signals under Risk Control Constraints]
    F2 --> G
    G --> H[Outcomes: Outperformed Buy & Hold Benchmark<br/>Fine-Tuned SLM Matched/Exceeded General LLM Performance]

Cited References (13)

1. TradingGroup: A Multi-Agent Trading System with Self-Reflection and Data-Synthesis (2025) — link
2. FinPos: A position-aware trading agent system for real financial markets (2025) — link
3. Trading-R1: Financial Trading with LLM Reasoning via Reinforcement Learning — → 2509.11420
4. FLAG-TRADER: Fusion LLM-Agent with Gradient-based Reinforcement Learning for Financial Trading — → 2502.11433
5. FinGPT: Open-Source Financial Large Language Models — → 2306.06031
6. BloombergGPT: A Large Language Model for Finance (2023) — link
7. PIXIU: A Large Language Model, Instruction Data and Evaluation Benchmark for Finance (2023) — link
8. FinAgent: A Multimodal Foundation Agent for Financial Trading: Tool-Augmented, Diversified, and Generalist — → 2402.18485

5 of 13 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.