基于SCD模型的金融超高频数据信息含量研究

针对随机波动条件持续期模型(SCD模型)对于金融超高频数据的模拟优势,采用股票交易的分笔数据,在模型中引入消息指标变量、交易量和买卖价差三类微观结构变量,通过分析价格持续期与这些结构变量之间的相关性,以研究价格形成过程中的信息传递过程。研究结果表明:利好消息通常导致交易强度增加,利空消息通常会导致交易持续期延长,价格变化对交易持续期造成的影响有限;小盘股买卖价差影响后续价格交易持续期,但大盘股的影响并不明显;交易量的增加具有信息含量。

Research on Financial UHF Data Information Content Based on SCD Model

In view of the simulation superiority of SCD model for financial UHF data, this paper, by using tick-by-tick data of the stock trading, adds the micro structure variables into the stochastic conditional duration model:message indicator variable, trading volume and bid-ask spread, and analyzes the relationship between price duration and these structure variables. The results show that good news led to increased trading and bad news led to decrease in trading, that price changes have limited impact on duration, that bid-ask spread of lag period is negatively related to the expect of price duration in next period and that the increased volume contains some news.