The riddle of volatility clusters

Abstract

In this financial engineering research we evaluate if observed non ­normalities in the market price distributions are caused mainly by a volatility clustering or also by another non­clustering mechanism. Such findings allow us to assess accor d­ ing to which rules the market price is actually developing or even make conclusions about market price directional forecasting chances, based on the realistic financial processes which we assign to the clustering and non ­clustering mechanisms. In the research we suggest certain methodology how to recognize these processes behind the market price development. We apply the method to the European government bonds market and for the comparison also to 1 day periods of S&P 500 Index development, with respect to the different time periods. Based on the findings we confirm the combination of both the volatility clustering and the non ­clustering processes to be active inside 1 day periods and to be responsible for measured non­normalities. We also find significant non ­clustering mechanism in 30 and 60 minute periods in case of European government bonds.