Statistical Behavior of the Nonlinear Correlation in Financial Markets

Abstract

We graft the volatility clustering in empirical financial time series into the Equiluz and Zimmermann model (EZ model), which was introduced to reproduce the herding behaviors of a financial time series. We examine the universality of the grafting methodology, which is one similar kind of sorting method that has been used to copy the nonlinear correlation structure of the real financial time series and to dress with it the model-based time series. According to the previous work, our grafting method proved that the nonlinear correlation structure of a herding model can be improved. In particular, we expand the same methodology to the high frequency financial time series and confirm that an improvement is achieved. Based on this result, we claim that a gap between real financial data and a model-based one can be narrowed.