基于小波分析的股指期货高频预测研究

基于低频金融数据的预测,在时间上具有长期性,依赖于整体经济环境,不能形成短期内的准确预测.但是由于高频金融时间序列具有非线性、非平稳性以及其特有的日历效应等特性,传统的ARMA模型也无法得到满意的预测结果.本文提出基于小波多分辨率分析的预测方法,将收益率数据分为高频部分(周期性)与低频部分(趋势性),对拆分后的序列进行重构,并对重构后得到的数据分别建立ARMA模型.实证研究表明,小波多分辨率分析能很好地滤出日内效应,由于股指期货独特的市场特征,应将分解层数定为3,分解重构模型可以提高预测精度.

Research on high frequency data forecasting of stock index futures market based on wavelet analysis

Prediction based on low-frequency financial data is long-term, depending on the overall economic environment, which is difficult to form accurate prediction. Because of the non-stationary, nonlinear and unique calendar effect of the financial high-frequency data, the traditional ARMA measuring method cannot get a satisfied analytical effect. We introduce a forecast method based on wavelet multi-resolution analysis, which can divide the yield data into high frequency part (periodicity) and low frequency part (tendency). We can reconstruct the separated sequence and make ARMA models. Results show wavelet multi-resolution analysis can filter the intraday effect well. Due to the unique characteristic of the stock index futures, the decomposed layer is 3. The empirical research shows that this reconstruct model improves the prediction precision.