中文摘要：

通过选取上海期货交易所燃油期货价格指数5分钟高频收益数据，本文构造了经调整的已实现波动率估计序列，运用4类非线性GARCH模型建模分析，描述了中国燃油期货价格指数的波动特征。运用6种损失函数以及Diebold—Mariano检验法，实证检验了4类GARCH模型对燃油期货价格指数波动的样本外预测能力。就中国燃油期货市场而言，基于高频数据的FIAPARCH模型，能够较好地描述中国燃油期货价格的波动特征，并且具有最为出色的波动率预测能力，而IGARCH模型在某些损失函数标准下也体现出了较好波动率预测能力。

英文摘要：

The present study selects high frequency data for 5 minutes of oil futures price index in the Shanghai Futures Exchange and constructs the adjusted realized volatility estimation sequence. It uses the modeling of 4 classes of nonlin- ear analysis of GARCH model and describes the fluctuation characteristics of Chinese fuel oil futures price index. Be- sides, it uses 6 kinds of loss function and the Diebold - Mariano inspection method to empirically test the ability of 4 class GARCH models on the fluctuation of fuel oil futures price index out of sample forecasting. As far as Chinese futures market is concerned, the FIAPARCH model based on high frequency data can better describe the volatility characteristics of Chinese fuel oil futures prices, and has the most excellent volatility forecasting ability. The IGARCI-I model also re- flects the good volatility forecasting ability in terms of some loss function standards.