TwinMarket: A Scalable Behavioral and Social Simulation for Financial Markets

TL;DR

TwinMarket is a scalable multi-agent framework utilizing large language models to simulate complex socio-economic phenomena like financial bubbles and recessions, capturing diverse human behaviors and emergent market dynamics.

Contribution

This work introduces TwinMarket, a novel LLM-based multi-agent simulation platform for modeling socio-economic systems and emergent financial phenomena.

Findings

LLM agents can simulate irrational behaviors and cognitive biases.

Emergent phenomena like bubbles and recessions arise from agent interactions.

The framework offers insights into collective socio-economic dynamics.

Abstract

The study of social emergence has long been a central focus in social science. Traditional modeling approaches, such as rule-based Agent-Based Models (ABMs), struggle to capture the diversity and complexity of human behavior, particularly the irrational factors emphasized in behavioral economics. Recently, large language model (LLM) agents have gained traction as simulation tools for modeling human behavior in social science and role-playing applications. Studies suggest that LLMs can account for cognitive biases, emotional fluctuations, and other non-rational influences, enabling more realistic simulations of socio-economic dynamics. In this work, we introduce TwinMarket, a novel multi-agent framework that leverages LLMs to simulate socio-economic systems. Specifically, we examine how individual behaviors, through interactions and feedback mechanisms, give rise to collective dynamics and emergent phenomena. Through experiments in a simulated stock market environment, we demonstrate how individual actions can trigger group behaviors, leading to emergent outcomes such as financial bubbles and recessions. Our approach provides valuable insights into the complex interplay between individual decision-making and collective socio-economic patterns.