The Research On Volatility Measurement And Modeling In China’s Stock Market Based On Realized Range Theory

Measurement and modeling of the volatility of financial assets is one of the core areas of financial economics.Select the optimum volatility estimator,and then conduct volatility modeling is the primary task of financial risk management.Proposed in recent years based on high frequency data,the realized range(RRV)compared to the realized volatility(RV)has better estimated volatility efficiency and attracts widespread attention.The proposition of realized range multipower variation further improved the estimated volatility system based on range.In the jump diffusion process,this paper applies Monte Carlo simulation data and high-frequency data on the Shanghai Composite Index to study the measurement of volatility、improved jump test statistic、volatility modeling and risk measurement based on realized range theory.The main work and conclusions are set out below:First,under the assumption of independent and identically distributed microstructure noise and independent with effective price,we conduct noise correction on realized range tripower variance,then compare the pros and cons of various types of volatility proxy variables through Monte Carlo simulation,concluding that the noise correction realized range tripower variance is the most effective,and lay a foundation for the next phase to improve jump test statistic and peel jump component.Second,for the local volatility estimated shortage in LM jump test statistic,use noise correction realized tripower variance to estimate local volatility instead of bipower variance(BV).Monte Carlo simulation results show that the improved LM jump test statistic has higher detection efficiency.Then apply it to analyze jump characteristics of China’s stock market.The results indicate the presence of asymmetric jumping behavior of China’s stock market,while jumping behavior has significant intraday pattern,but weeks mode is not obvious.Third,with reference to the quadratic variation theory,this paper peels jump component according to the improved LM jump test results.Decompose the realized range into continuous sample path variance and jump variance.Considering the continuous portion、jump portion and leverage effect to build HAR-RRV type volatility model.In-sample and out-of-sample forecast results both show LHAR-RRV-CJ model performs best.Finally,based on the out-of-sample forecast volatility to estimate ES risk measurement.The results show that optimum volatility forecasting model also has better risk measurement effect.Skew t distribution is better than normal distribution on characterizing China’s stock market returns.