中文摘要：

提出了基于分类信息的C-GARCH模型和S-GARCH模型,并结合传统未考虑分类信息下的GARCH模型,以上证综指五分钟数据为样本,对波动率进行了实证分析;研究结果表明：分类信息GARCH模型优于未考虑分类信息的模型,最优模型为C-GARCH模型,其次为S-GARCH模型;好消息和坏消息对高频数据方差的影响程度相对较小,但却提高了描述精度;好消息与方差波动负相关,坏消息与方差波动正相关;坏消息对波动率的影响比好消息大,具有非对称性。

英文摘要：

C-GARCH Model and S-GARCH Model based on classification information are proposed by combining traditional GARCH Model without considering classification information. Taking Shanghai Composite Index in five minutes as an example, empirical analysis is conducted on its volatility rate. Research results show that classification information GARCH Model is better than the Model without considering classification information, that C-GARCH Model is the optimal model and S-GARCH Model is the second, that the influence degree of good news and bad news on high frequency data variance is small but improves its descriptive accuracy, that good news is negatively related to variance volatility but bad news is positively related to variance volatility and that the influence of bad news on volatility rate is bigger than that of good news and is asymmetric.