Papers
Topics
Authors
Recent
Search
2000 character limit reached

A Multimodal Foundation Agent for Financial Trading: Tool-Augmented, Diversified, and Generalist

Published 28 Feb 2024 in q-fin.TR and cs.AI | (2402.18485v3)

Abstract: Financial trading is a crucial component of the markets, informed by a multimodal information landscape encompassing news, prices, and Kline charts, and encompasses diverse tasks such as quantitative trading and high-frequency trading with various assets. While advanced AI techniques like deep learning and reinforcement learning are extensively utilized in finance, their application in financial trading tasks often faces challenges due to inadequate handling of multimodal data and limited generalizability across various tasks. To address these challenges, we present FinAgent, a multimodal foundational agent with tool augmentation for financial trading. FinAgent's market intelligence module processes a diverse range of data-numerical, textual, and visual-to accurately analyze the financial market. Its unique dual-level reflection module not only enables rapid adaptation to market dynamics but also incorporates a diversified memory retrieval system, enhancing the agent's ability to learn from historical data and improve decision-making processes. The agent's emphasis on reasoning for actions fosters trust in its financial decisions. Moreover, FinAgent integrates established trading strategies and expert insights, ensuring that its trading approaches are both data-driven and rooted in sound financial principles. With comprehensive experiments on 6 financial datasets, including stocks and Crypto, FinAgent significantly outperforms 9 state-of-the-art baselines in terms of 6 financial metrics with over 36% average improvement on profit. Specifically, a 92.27% return (a 84.39% relative improvement) is achieved on one dataset. Notably, FinAgent is the first advanced multimodal foundation agent designed for financial trading tasks.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (57)
  1. Optuna: A Next-Generation Hyperparameter Optimization Framework. In The 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. 2623–2631.
  2. Deep reinforcement learning for quantitative trading: Challenges and opportunities. IEEE Intelligent Systems 37, 2 (2022), 23–26.
  3. Investment behaviors can tell what inside: Exploring stock intrinsic properties for stock trend prediction. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. 2376–2384.
  4. Theoremqa: A theorem-driven question answering dataset. arXiv preprint arXiv:2305.12524 (2023).
  5. Palm: Scaling language modeling with pathways. Journal of Machine Learning Research 24, 240 (2023), 1–113.
  6. Pangu-Agent: A fine-tunable generalist agent with structured reasoning. arXiv:2312.14878 [cs.AI]
  7. Deep direct reinforcement learning for financial signal representation and trading. IEEE Transactions on Neural Networks and Learning Systems 28, 3 (2016), 653–664.
  8. Investor-imitator: A framework for trading knowledge extraction. In Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. 1310–1319.
  9. Technical analysis of stock trends. CRC press.
  10. Temporal relational ranking for stock prediction. ACM Transactions on Information Systems (TOIS) 37, 2 (2019), 1–30.
  11. Soft actor-critic algorithms and applications. arXiv preprint arXiv:1812.05905 (2018).
  12. MetaGPT: Meta programming for a multi-agent collaborative framework. arXiv:2308.00352 [cs.AI]
  13. Listening to chaotic whispers: A deep learning framework for news-oriented stock trend prediction. In Proceedings of the 11th ACM International Conference on Web Search and Data Mining (WSDM). 261–269.
  14. Billion-scale similarity search with gpus. IEEE Transactions on Big Data 7, 3 (2019), 535–547.
  15. Lightgbm: A highly efficient gradient boosting decision tree. Advances in neural information processing systems 30 (2017).
  16. FinRL: A deep reinforcement learning library for automated stock trading in quantitative finance. Deep RL Workshop, NeurIPS 2020 (2020).
  17. Adaptive quantitative trading: An imitative deep reinforcement learning approach. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 34. 2128–2135.
  18. Mathvista: Evaluating mathematical reasoning of foundation models in visual contexts. arXiv preprint arXiv:2310.02255 (2023).
  19. Learn to explain: Multimodal reasoning via thought chains for science question answering. Advances in Neural Information Processing Systems 35 (2022), 2507–2521.
  20. Chameleon: Plug-and-Play compositional reasoning with large language models. arXiv:2304.09842 [cs.CL]
  21. Dynamic prompt learning via policy gradient for semi-structured mathematical reasoning. arXiv preprint arXiv:2209.14610 (2022).
  22. Playing atari with deep reinforcement learning. arXiv preprint arXiv:1312.5602 (2013).
  23. Webgpt: Browser-assisted question-answering with human feedback. arXiv preprint arXiv:2112.09332 (2021).
  24. Stock market’s price movement prediction with LSTM neural networks. In 2017 International Joint Conference on Neural Networks (IJCNN). 1419–1426.
  25. OpenAI. 2021. Chatgpt. https://openai.com/research/chatgpt
  26. OpenAI. 2023a. GPT-4 Technical Report. arXiv:2303.08774 [cs.AI]
  27. OpenAI. 2023b. GPT-4V(ision) system card. https://openai.com/research/gpt-4v-system-card
  28. Talm: Tool augmented language models. arXiv preprint arXiv:2205.12255 (2022).
  29. Generative Agents: Interactive simulacra of human behavior. arXiv:2304.03442 [cs.HC]
  30. Earnhft: Efficient hierarchical reinforcement learning for high frequency trading. arXiv preprint arXiv:2309.12891 (2023).
  31. Deep attentive learning for stock movement prediction from social media text and company correlations. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP). 8415–8426.
  32. VolTAGE: Volatility forecasting via text-audio fusion with graph convolution networks for earnings calls. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP). 8001–8013.
  33. Quantitative day trading from natural language using reinforcement learning. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 4018–4030.
  34. Toolformer: Language models can teach themselves to use tools. arXiv preprint arXiv:2302.04761 (2023).
  35. Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347 (2017).
  36. Cognitive architectures for language agents. arXiv preprint arXiv:2309.02427 (2023).
  37. Scieval: A multi-level large language model evaluation benchmark for scientific research. arXiv preprint arXiv:2308.13149 (2023).
  38. PRUDEX-Compass: Towards systematic evaluation of reinforcement learning in financial markets. Transactions on Machine Learning Research (2023).
  39. TradeMaster: A holistic quantitative trading platform empowered by reinforcement learning. In Thirty-seventh Conference on Neural Information Processing Systems Datasets and Benchmarks Track.
  40. Reinforcement learning for quantitative trading. ACM Transactions on Intelligent Systems and Technology 14, 3 (2023), 1–29.
  41. Mastering stock markets with efficient mixture of diversified trading experts. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’23). 2109–2119.
  42. Lamda: Language models for dialog applications. arXiv preprint arXiv:2201.08239 (2022).
  43. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971 (2023).
  44. Voyager: An open-ended embodied agent with large language models. arXiv preprint arXiv: Arxiv-2305.16291 (2023).
  45. CLVSA: A convolutional LSTM based variational sequence-to-sequence model with attention for predicting trends of financial markets. In Proceedings of the 28th International Joint Conference on Artificial Intelligence (IJCAI). 3705–3711.
  46. Commission fee is not enough: A hierarchical reinforced framework for portfolio management. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 35. 626–633.
  47. Scibench: Evaluating college-level scientific problem-solving abilities of large language models. arXiv preprint arXiv:2307.10635 (2023).
  48. DeepTrader: a deep reinforcement learning approach for risk-return balanced portfolio management with market conditions Embedding. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 35. 643–650.
  49. Bloomberggpt: A large language model for finance. arXiv preprint arXiv:2303.17564 (2023).
  50. Yumo Xu and Shay B Cohen. 2018. Stock movement prediction from tweets and historical prices. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (ACL). 1970–1979.
  51. FinGPT: Open-Source Financial Large Language Models. arXiv preprint arXiv:2306.06031 (2023).
  52. Auto-GPT for Online Decision Making: Benchmarks and Additional Opinions. arXiv:2306.02224 [cs.AI]
  53. MM-REACT: Prompting ChatGPT for multimodal reasoning and action. arXiv:2303.11381 [cs.CV]
  54. Reinforcement-learning based portfolio management with augmented asset movement prediction states. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 34. 1112–1119.
  55. Generate rather than retrieve: Large language models are strong context generators. arXiv preprint arXiv:2209.10063 (2022).
  56. FinMem: A performance-enhanced LLM trading agent with layered memory and character design. arXiv:2311.13743 [q-fin.CP]
  57. AppAgent: Multimodal agents as smartphone users. arXiv:2312.13771 [cs.CV]
Citations (13)
View on Semantic Scholar

Summary

  • The paper introduces FinAgent, which integrates multimodal data and a reflection module to enhance trading decisions.
  • The methodology employs market intelligence, memory, and decision-making modules while outperforming state-of-the-art methods with over 36% profit improvement.
  • Experimental results demonstrate FinAgent's ability to adapt to dynamic market conditions, highlighting its potential for broader financial applications.

A Multimodal Foundation Agent for Financial Trading: Tool-Augmented, Diversified, and Generalist

Introduction

The paper introduces FinAgent, a multimodal foundational agent designed specifically for financial trading. FinAgent tackles the challenges encountered in financial trading applications, which include inadequate handling of multimodal data and limited generalizability across various tasks. By integrating diverse data types and employing a comprehensive reflection module, FinAgent aims to improve adaptability and decision-making in dynamic market environments. Figure 1

Figure 1: The overall architecture of FinAgent. The ordinal numbers in the figure represent the order of execution, where augmented tools are implemented with the decision-making module.

Architecture and Modules of FinAgent

FinAgent is structured around a modular architecture that incorporates market intelligence, memory, reflection, and decision-making components:

  1. Market Intelligence Module: This module processes numerical, textual, and visual data to analyze market trends. It employs diversified retrieval operations to gather comprehensive insights from historical data, which are then summarized to inform decision-making tasks.
  2. Memory Module: Utilizing a vector storage architecture, this module supports information retrieval and storage for market intelligence and reflection modules. It offers functionalities crucial for acuity, adaptability, and amendability in decision-making.
  3. Reflection Module: Comprised of low-level and high-level reflection components, this module examines the relationships between market observations and price changes as well as assesses past trading decisions. These reflections foster learning and help in refining strategies.
  4. Decision-Making Module: Incorporating tool-augmented models, this module synthesizes insights from the other modules along with expert knowledge to make final trading decisions. The decision-making process emphasizes reasoning and integrates auxiliary trading tools.

Implementation and Performance

FinAgent has been evaluated on six diverse financial datasets, including both stock and cryptocurrency markets. It outperformed nine state-of-the-art baseline methods across six financial metrics, achieving an average profit improvement of over 36% and attaining a notable 92.27% return on one dataset. Figure 2

Figure 2: Case studies of FinAgent. We only display the partial prompt for brevity. See Appendix for the full prompt structure.

Diversified Retrieval and Reflection in FinAgent

FinAgent's diversified retrieval approach enables more targeted and noise-resistant information gathering, enriching the agent's market insights. The reflection module's inclusion of both low-level and high-level reflections allows it to adapt learning processes akin to human reasoning, enhancing decision outcomes based on comprehensive evaluations of historical and current data. Figure 3

Figure 3: Performance comparison over time between FinAgent and other benchmarks across all assets.

Experimental Insights

The experimental evaluation revealed that FinAgent consistently identifies profitable trading opportunities by leveraging its modular framework. It excels in environments where multimodal data integration and reasoning-based approaches are critical. However, it was observed that in cryptocurrency markets, supplementary trading strategies need to be more generalized for improved performance.

Conclusion

The introduction of FinAgent marks a significant advancement in autonomous trading systems, blending multimodal data processing with cognitive reflection and tool augmentation. Future developments could focus on generalizing auxiliary agents for broader market applicability and exploring other financial applications beyond trading tasks.

File Document Download Save Streamline Icon: https://streamlinehq.com Markdown

Paper to Video (Beta)

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Sign Up to Generate

Open Problems

We found no open problems mentioned in this paper.

Collections

Sign up for free to add this paper to one or more collections.

User Circle Single Streamline Icon: https://streamlinehq.com Sign Up

Tweets

Sign up for free to view the 8 tweets with 2 likes about this paper.

User Circle Single Streamline Icon: https://streamlinehq.com Sign Up for Free