Analysis of Spin Financial Market by GARCH Model

TL;DR

This paper uses a GARCH model with Bayesian inference to analyze simulated spin financial markets, demonstrating volatility clustering and supporting the mixture-of-distribution hypothesis for asset returns.

Contribution

It applies Bayesian MCMC to estimate GARCH parameters in a spin model simulation, linking simulated market volatility to real market phenomena.

Findings

Volatility exhibits clustering similar to real markets

Standardized returns are approximately normal

No significant autocorrelation in absolute standardized returns

Abstract

A spin model is used for simulations of financial markets. To determine return volatility in the spin financial market we use the GARCH model often used for volatility estimation in empirical finance. We apply the Bayesian inference performed by the Markov Chain Monte Carlo method to the parameter estimation of the GARCH model. It is found that volatility determined by the GARCH model exhibits "volatility clustering" also observed in the real financial markets. Using volatility determined by the GARCH model we examine the mixture-of-distribution hypothesis (MDH) suggested for the asset return dynamics. We find that the returns standardized by volatility are approximately standard normal random variables. Moreover we find that the absolute standardized returns show no significant autocorrelation. These findings are consistent with the view of the MDH for the return dynamics.