基于序列比对方法的股市波动实证研究

Empirical Research of Stock Market Volatility Based on Sequence Alignment Method

引入生物信息学中的序列比对方法及非参数的符号时间序列分析方法,与已有的K-近邻法相结合,提出一种新的股价波动预测方法。将样本序列与比对目标序列进行全局比对,通过动态规划算法回溯出高于匹配得分阈值的K条历史子序列,以此作为K-近邻法中的K个最近邻,从而得到预测结果。该方法不仅可以预测具体的波动值,也可预测波动所处的区间。以上证综指和深证成指采样间隔为20 min的高频数据为样本进行实证分析,验证了该方法的有效性。

Through the introduction of sequence alignment method in bioinformatics and the nonparametric symbolic time series analysis,a new method of forecasting stock price volatility was put forward combined with the existing K-Nearest Neighbor method.Global Comparison was made between sample series and the alignment goal sequence.K historical sub-sequences whose matching score are higher than the threshold value were backed out using the dynamic programming algorithm and viewed as K nearest neighbor in K-NN method to get the predicting outcomes.Not only the specific fluctuation but also the fluctuating interval can be predicted through this method.Empirical research was done with high frequency data whose sampling interval is 20 minutes from Shanghai Stock Exchange Composite Index and Shenzhen Stock Exchange Component Index.The validity of this method was proved.