A Research On Volatility And The Application Of Forecasting Volatility Model In Chinese Soybean Futures Market

In recent years,the dependence on the import of soybeans gets more serious in our country.The variety of the soybean's price have an important impact on the entire soybean industry and the national economy.The soybean futures are the core of the soybean market mechanism,which is very active in agricultural futures,so this article takes soybean futures as a point to study its volatility.We select the high frequency data and the daily trading data of soybean futures as the study samples.This paper takes the mainstream GARCH-type model,the common realized volatility models and models which are revolved on their basis to fit its in-sample data and predict its volatility.To model the GARCH(1,1)and test its asymmetric effect are found that the soybean futures does not exist the asymmetric effect which are existed widely in stock markets.From the new information curve,we can also know that the responses of the soybean futures market to the positive and negative news are consistent and symmetrical.By establishing the HAR-RV model,we found that soybean futures have obvious long memory characteristics,which show that it is feasible to estimate the future volatility through the historical information of soybean futures.Finally,the rolling time windows one-day-ahead forecasting method are used to predict the future volatility of various volatility models.Then,six kind of loss functions are used as the criterion of the model assessment,and the prediction accuracy of the model is carried out by the latest MCS test.It is found that the daily trading volume and RV as an additional explanatory variable greatly improve the prediction accuracy of the original model and are obviously superior to the classical GARCH family and the partially RV model.This research has shown that GARCH(1,1)-RV and ARFIMA-RV are the best performer in all these forecast models in the soybean futures market.