Leveraging Large Language Models for Institutional Portfolio Management: Persona-Based Ensembles

Yoshia Abe,Shuhei Matsuo,Ryoma Kondo,Ryohei Hisano

2024-11-29

Abstract:Large language models (LLMs) have demonstrated promising performance in various financial applications, though their potential in complex investment strategies remains underexplored. To address this gap, we investigate how LLMs can predict price movements in stock and bond portfolios using economic indicators, enabling portfolio adjustments akin to those employed by institutional investors. Additionally, we explore the impact of incorporating different personas within LLMs, using an ensemble approach to leverage their diverse predictions. Our findings show that LLM-based strategies, especially when combined with the mode ensemble, outperform the buy-and-hold strategy in terms of Sharpe ratio during periods of rising consumer price index (CPI). However, traditional strategies are more effective during declining CPI trends or sharp market downturns. These results suggest that while LLMs can enhance portfolio management, they may require complementary strategies to optimize performance across varying market conditions.

Computational Engineering, Finance, and Science,Multiagent Systems

What problem does this paper attempt to address?

The problems that this paper attempts to solve are: How to use large - language models (LLMs) to predict price movements in stock and government bond portfolios, and adjust positions based on these predictions to achieve portfolio management similar to that of institutional investors. In addition, the research also explores the impact of incorporating different personas in LLMs on prediction accuracy, and improves prediction performance through the ensemble approach.

Specifically, the main objectives of the paper include:
1. Design prompts that enable LLMs to conduct portfolio management according to the settings of institutional investors.
2. Research the differences in the performance of strategies based on LLMs' personas, and utilize them through the ensemble approach.
3. Quantitatively analyze the periods when LLMs' strategies perform excellently, and qualitatively analyze the prediction reasons for each persona generated by LLMs.
4. Demonstrate that during the upward trend of the consumer price index (CPI), the LLMs - based strategies are superior to traditional methods in terms of the Sharpe ratio.

Through these objectives, the paper aims to explore the application potential of LLMs in complex investment strategies, especially for the portfolio management of institutional investors.