论文部分内容阅读
以我国期货市场上交易最为活跃的沪深300股指期货为例,分别采用CAViaR模型和GARCH模型对多头VaR和空头VaR进行风险建模,深入研究了股指期货的收益分布特征和波动形态规律,并运用严谨的后测检验的方法对比了各个模型的风险预测精度。实证结果表明:(1)沪深300股指期货具有明显的“尖峰厚尾”现象,却没有显著的有偏性和长记忆性;(2)基于杠杆效应的GJR模型和兼具长记忆性和杠杆效应的FIAPARCH模型并没有表现出比传统GARCH模型更高的预测精度,同时,先验GED分布对金融收益分布特征的刻画要优于正态分布和SKST分布;(3)半参数法的CAViaR模型相比GARCH族模型表现出绝对优异的预测能力。总之,CAViaR模型在股指期货的风险预测方面是相对更合理的模型选择。
Take the CSI 300 stock index futures which are the most traded on the futures market of our country as an example, we use the CAViaR model and the GARCH model respectively to model the long VaR and the short VaR, and deeply study the return distribution and fluctuation patterns of stock index futures. The rigorous post-test method is used to compare the risk prediction accuracy of each model. The empirical results show that: (1) the Shanghai and Shenzhen 300 stock index futures have obvious “thick tail ” phenomenon, but there is no significant bias and long memory; (2) the leverage based GJR model and both long memory The FIAPARCH model does not show a higher prediction accuracy than the traditional GARCH model, meanwhile, the prior GED distribution is better than the normal distribution and the SKST distribution in the distribution of financial returns. (3) Semi-parametric method The CAViaR model shows absolutely superior predictive power compared to the GARCH family model. In conclusion, the CAViaR model is a relatively more reasonable model choice in the risk prediction of stock index futures.