Agent-based model with asymmetric trading and herding for complex financial systems

Area of Science:

• Quantitative Finance
• Computational Economics
• Agent-Based Modeling

Background:

• Understanding return-volatility correlations (leverage and anti-leverage effects) is crucial for financial price dynamics.
• The microscopic origins of these effects and their simulation in agent-based models remain poorly understood.
• Determining model parameters from empirical data offers a promising alternative to traditional statistical fitting.

Purpose of the Study:

• To investigate the microscopic origins of return-volatility correlations in financial systems.
• To develop an agent-based model incorporating novel mechanisms for simulating these effects.
• To establish methods for parameter determination using historical market data.

Main Methods:

• Constructed an agent-based model simulating individual and collective investor behavior.
• Introduced asymmetric trading and herding mechanisms specific to bull and bear markets.
• Developed methods to determine key model parameters from historical stock market data.

Main Results:

• Simulations using parameters from six global stock indices reproduced empirical leverage/anti-leverage effects in amplitude and duration.
• The model successfully replicated other real market features, including fat-tailed return distributions and long-term volatility correlations.
• Investor trading asymmetry and herding were identified as essential mechanisms for generating these effects.

Conclusions:

• Asymmetric trading and herding are key drivers of leverage and anti-leverage effects.
• In markets exhibiting leverage effects, symmetric trading contributed minimally, highlighting the role of asymmetry.
• The developed mechanisms and parameter determination methods are applicable to other complex systems with asymmetries.