Assessment of the Impact and Mechanism of Smart Money Traders through Agent-based Market Simulations

Abstract

In financial markets, the term "Smart money" denotes big, experienced, and well-informed investors who deeply understand the market. As these investors are known for their strategic approach, long-term perspective, access to valuable information, and large capital, their decisions often yield substantial impacts, capable of shaping market sentiment and direction. Analyzing the behaviors of smart money traders can provide valuable insights for individual investors and traders. In this study, we explore the effect of smart money traders in the market by using artificial market simulation. We revise the existing model for artificial market simulation by introducing the new smart money agents and technical analysis agents and adjusting the mechanism to match with the conditions of different types of markets. Smart money agents possess advantageous capital and information sources, enabling them to make orders that significantly impact price movements and gain access to market changes promptly during periods of stress. These agents are trained with the deep reinforcement learning algorithm, while technical analysis agents employ investment strategies popular in real markets. We also investigate the conditions and compositions of the agents that lead to statistical properties similar to the features of real markets.