The Comparison Of The Forecasting Models Of Volatility

Volatility is one of financial variables with high degree of concern. Moreover, volatility forecasting is an important element in the research of volatility. Generally, volatility forecasting model can be divided into two categories. The historical volatility method is one the methods to forecast the future volatility by using the historical transaction information in the market; and deterministic volatility models, time serials models and stochastic models are included in this method. Another method is called implied volatility method. By using this method, researchers can have expects of future volatility through the price of options in the market. This method contains two calculation methods which are BS Implied Volatility and Model-Free Implied Volatility.The main purpose of this paper is to compare the different prediction models, and the Model-Free Implied Volatility which is less used in previous studies will be included in the scope of the comparison. Compare with the historical volatility method, the Model-Free Implied Volatility has more market information. Moreover, implied volatility method will be divided into BS Implied Volatility and Model-Free Implied Volatility while comparing. And compare with BS Implied Volatility, the Model-Free Implied Volatility is not require to use the BS pricing formula, and it doesn't have to face the potential risks.In 50 ETF option market, the forecasting ability of GARCH model, weighted BS Implied Volatility and free Model-Free Implied Volatility have been compared and researched through comparing information content. The conclusion shows, the three methods above cannot forecast the future volatility accurately when securities market is unstable. On the other hand, the volatility in securities market can be forecasted in stable period. Refers to the conclusion, when the forecasting time was limited in a month, GARCH model has worst forecasting ability due to the fewest information content and free model's ability of forecasting is better than BS Implied Volatility because of more information content.