Study Of Risk Measures On CSI 300 Index Based On Realized GARCH Model

Study on volatility has important meaning in the field of risk management,hedging and assets pricing.With the development of computer science and data storage,it becomes possible to adopt high-frequency data to build volatility model.This paper builds Realized GARCH model using 5 minutes high-frequency data based on CSI 300 index to fit volatility as well as measure risk,which considering of different realized measures,residual distribution and non-trading return factors.Realized Variance,Realized Range and two kinds of adjust RV considering of overnight return and lunchtime break are choose as the realized measures.Assuming residual is respectively subject to norm distribution,Student t distribution,GH distribution and GHST distribution.At last,compare the difference between daily return based on open-close return and close-close return.The empirical results show that the Realized GARCH model adopting RR and GH distribution which based on open-close return has the best performance on fitting volatility.In the meantime,changing realized measures has the greatest influence on fitting effect among all these three factors.In the field of risk measurement,that is calculating VaR and ES,realized measures' impact is negligible.On the other hand,residual that subjects to fat-tail distribution performs better when measuring extreme risk at the tail.Advantage of applying open-to-close return in the field of fitting volatility continued in the field of risk measurement.