Paper: arXiv 2412.20138
Authors: Yijia Xiao, Edward Sun, Di Luo, Wei Wang
Abstract
Significant progress has been made in automated problem-solving using societies of agents powered by large language models (LLMs). In finance, efforts have largely focused on single-agent systems handling specific tasks or multi-agent frameworks independently gathering data. However, the multi-agent systems’ potential to replicate real-world trading firms’ collaborative dynamics remains underexplored. TradingAgents proposes a novel stock trading framework inspired by trading firms, featuring LLM-powered agents in specialized roles such as fundamental analysts, sentiment analysts, technical analysts, and traders with varied risk profiles. The framework includes Bull and Bear researcher agents assessing market conditions, a risk management team monitoring exposure, and traders synthesizing insights from debates and historical data to make informed decisions. By simulating a dynamic, collaborative trading environment, this framework aims to improve trading performance. Detailed architecture and extensive experiments reveal its superiority over baseline models, with notable improvements in cumulative returns, Sharpe ratio, and maximum drawdown, highlighting the potential of multi-agent LLM frameworks in financial trading. TradingAgents is available at https://github.com/TauricResearch/TradingAgents.
Complexity vs Empirical Score
- Math Complexity: 2.5/10
- Empirical Rigor: 6.0/10 (capped from 8.0: transaction costs not modelled; see penalty flags)
- Quadrant: Street Traders — empirical and applied, lower math complexity (effective empirical rigor capped from 8.0)
Why this score: The paper focuses on LLM agent architecture and backtesting metrics like Sharpe ratio and cumulative returns, with minimal advanced mathematics beyond basic statistics, making it highly empirical and implementation-focused.
Research Flowchart
flowchart TD A["Research Goal<br>Explore multi-agent LLM framework<br>to replicate trading firm dynamics"] --> B["Methodology<br>Multi-agent system with specialized<br>roles & collaborative debates"] B --> C["Data/Inputs<br>Financial market data &<br>historical trading information"] C --> D["Computational Process<br>Agent debates & synthesis of<br>fundamental, technical & sentiment insights"] D --> E["Key Findings<br>Superior performance vs baselines<br>↑ Cumulative returns, ↑ Sharpe ratio,<br>↓ Maximum drawdown"] E --> F["Outcome<br>Validates multi-agent LLM<br>potential in financial trading"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes TradingAgents, a novel multi-agent LLM-powered stock trading framework with specialized analyst and trader roles.
Source sentence
TradingAgents proposes a novel stock trading framework inspired by trading firms, featuring LLM-powered agents in specialized roles such as fundamental analysts, sentiment analysts, technical analysts, and traders with varied risk profiles.
Assumes 1
- The framework assumes the existence of defined teams of specialized agents (analyst team, researcher team, trader team, risk manager). model
Source sentence
More specifically, TradingAgents assumes the following teams.
Data
2024-01 – 2024-03 · US equities · multi-asset · daily
- Multi-asset multi-modal financial dataset other unknown
Source sentence
Our multi-modal dataset integrates historical stock prices, news articles, social media sentiment, insider transactions, financial statements, and 60 technical indicators per asset.
Costs
Not modeled
Finds 2
annual return 24.9% · total return 23.2% · max drawdown 2.1%
- TradingAgents outperforms baseline strategies in cumulative return, Sharpe ratio, and maximum drawdown. performance
Source sentence
Experiments demonstrate that TradingAgents outperforms traditional trading strategies and baselines in cumulative return, Sharpe ratio, and other critical financial metrics.
- The framework shows superiority over baseline models with notable improvements in cumulative returns, Sharpe ratio, and maximum drawdown. performance
Source sentence
Detailed architecture and extensive experiments reveal its superiority over baseline models, with notable improvements in cumulative returns, Sharpe ratio, and maximum drawdown
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 30 related papers). Coverage is partial, so this is neither replication nor refutation.
38 papers in the archive : Agent Policy-Value Audit: Separating Transition Composition … ; FinRL-X: An AI-Native Modular Infrastructure for …
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (45)
- FinCon: A Synthesized LLM Multi-Agent System with Conceptual Verbal Reinforcement for Enhanced Financial Decision Making (2024) — link
- TradingGPT: Multi-agent system with layered memory and distinct characters for enhanced financial trading performance — → 2309.03736
- QuantAgent: Seeking Holy Grail in Trading by Self-Improving Large Language Model — → 2402.03755
- Large Language Model Agent in Financial Trading: A Survey — → 2408.06361
- ReAct: Synergizing Reasoning and Acting in Language Models (2023) — link
- Can Large Language Models Beat Wall Street? Unveiling the Potential of AI in Stock Selection — → 2401.03737
- FinRobot: An Open-Source AI Agent Platform for Financial Applications using Large Language Models — → 2405.14767
- Auto-GPT for Online Decision Making: Benchmarks and Additional Opinions (2023) — link
and 37 more
- MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework (2024) — link
- Improving Factuality and Reasoning in Language Models through Multiagent Debate (2023) — link
- Evaluation of OpenAI o1: Opportunities and Challenges of AGI (2024) — link
- On The Planning Abilities of OpenAI’s o1 Models: Feasibility, Optimality, and Generalizability (2024) — link
- Learning to Reason with LLMs - OpenAI O1 Model (2024)
- GPT-4 Technical Report (2024) — link
- Generative Agents: Interactive Simulacra of Human Behavior (2023) — link
- ChatDev: Communicative Agents for Software Development (2024) — link
- Teaching Large Language Models to Reason with Reinforcement Learning (2024) — link
- Multi-Agent Collaboration: Harnessing the Power of Intelligent LLM Agents (2023) — link
- MedAgents: Large Language Models as Collaborators for Zero-shot Medical Reasoning (2024) — link
- PIXIU: A Large Language Model, Instruction Data and Evaluation Benchmark for Finance (2023) — link
- FinGPT: Open-Source Financial Large Language Models — → 2306.06031
- Instruct-FinGPT: Financial Sentiment Analysis by Instruction Tuning of General-Purpose Large Language Models — → 2306.12659
- OPT: Open Pre-trained Transformer Language Models (2022) — link
- BloombergGPT: A Large Language Model for Finance (2023) — link
- XuanYuan 2.0: A Large Chinese Financial Chat Model with Hundreds of Billions Parameters (2023) — link
- BBT-Fin: Comprehensive Construction of Chinese Financial Domain Pre-trained Language Model, Corpus and Benchmark (2023) — link
- PaLM: Scaling Language Modeling with Pathways (2022) — link
- Llama 2: Open Foundation and Fine-Tuned Chat Models (2023) — link
- Qwen Technical Report (2023) — link
- Baichuan 2: Open Large-scale Language Models (2023) — link
- Can ChatGPT Forecast Stock Price Movements? Return Predictability and Large Language Models — → 2304.07619
- Unveiling the Potential of Sentiment: Can Large Language Models Predict Chinese Stock Price Movements? — → 2306.14222
- Sentiment trading with large language models — → 2412.19245
- Can Large Language Models Beat Wall Street? Unveiling the Potential of AI in Stock Selection — → 2401.03737
- LLMFactor: Extracting Profitable Factors through Prompts for Explainable Stock Movement Prediction (2024) — link
- FinMem: A Performance-Enhanced LLM Trading Agent with Layered Memory and Character Design — → 2311.13743
- A Multimodal Foundation Agent for Financial Trading: Tool-Augmented, Diversified, and Generalist — → 2402.18485
- Memory: enduring traces of perceptual and reflective attention (2011) — link
- Towards Mitigating Hallucination in Large Language Models via Self-Reflection (2023) — link
- Designing Heterogeneous LLM Agents for Financial Sentiment Analysis (2024) — link
- TradingGPT: Multi-Agent System with Layered Memory and Distinct Characters for Enhanced Financial Trading Performance — → 2309.03736
- Alpha-gpt: Human-ai interactive alpha mining for quantitative investment — → 2308.00016
- Learning to Generate Explainable Stock Predictions using Self-Reflective Large Language Models — → 2402.03659
- Integrating Stock Features and Global Information via Large Language Models for Enhanced Stock Return Prediction — → 2310.05627
- Proximal Policy Optimization Algorithms (2017) — link
17 of 45 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.