An Analysis On The Forecasting Effect Of Stock Index Volatility On The Return Of Index Fund

Volatility is a very important variable in the study of modern financial economics.It is widely used in the investment,pricing,risk control and monetary policy making.Therefore,in the recent 20 years,many scholars and professionals use different Using different methods to make different estimates and forecasts of the volatility and get different conclusions.The existence of many models also shows its own characteristics.From the existing research,the study of volatility from different aspects usually leads to different conclusions,which in turn shows that all kinds of models have different defects.For the industry,the estimation and prediction of the volatility are directly related to the advantages and disadvantages of the construction of the investment portfolio and have a direct effect on the effectiveness of the risk management.Therefore,studying the volatility has been a key issue in academia and industry.These models are mainly divided into two categories: one is to estimate the volatility through the historical price of the underlying asset,referred to as the historical information method for short,such as the GARCH model,the Stochastic Volatility Model(SV)and the Realized Volatility Model(RV).The other method is essentially different from the historical information method.It uses the connotation information of the option price to extract the volatility of the underlying asset from the option price,namely the Implied Volatility(IV).This article chooses random volatility,realized volatility and model-free implied volatility to study the full text.The reasons for choosing these three kinds of volatility are as follows: Firstly,we use these three kinds of volatility to study the influence of volatility under different measurement ranges on the index return.Stochastic volatility and realized volatility are the volatility obtained from the underlying asset price and can be considered as the real measure volatility.The implied volatility is the volatility derived from the option price.The option pricing is based on the risk-neutral world So implied volatility can be considered as the volatility of a risk-neutral world.Second,we want to study the effect of different data frequencies on the volatility forecasting effect.Stochastic volatility is the use of daily data to estimate the volatility,but the volatility is the use of intraday high frequency data to estimate the volatility,so the data frequency of the volatility of the estimated effect is also an important issue in this study.In this paper,different volatility estimation results are obtained based on the three volatility estimation methods,and the volatility estimation results under three different methods are used to predict the index fund yield.In the forecast content,the paper uses the volatility obtained under three different methods to forecast daily yield,weekly rate of return and monthly rate of return of Shanghai 50 ETF index respectively.In the prediction method,this paper makes use of the univariate regression model and the state space model,Respectively,the constant coefficient and time-varying coefficient for Shanghai 50 ETF index yield forecast.The empirical results show that there is a strong correlation between the three volatility,indicating that although the volatility estimates are inconsistent,but the inherent meaning of the reflection is the same;the use of the volatility of the three models estimated the rate of return When forecasting,the overall forecasting effect of stochastic volatility is the best,and the forecasting effect of model-less implied volatility is the second,and the forecasting effect of realized volatility is the worst.Compared with daily returns,The weekly yield and the monthly yield have a better forecasting effect,ie,the longer the yield period,the more significant the forecasting effect is.Relative to the one-dimensional regression forecasting model,the time-varying parameter forecasting model obtained from the state space model can be used to predict the index return The effect is significantly better than the one-way regression method to predict the effect.