中文摘要：

为了准确的把握股价的趋势走向，提出了一种基于极大重叠离散小波变换（MODWT）时间序列分析的股价预测方法（MARMA）。该方法是对股价时间序列利用mallat算法对其进行极大重叠离散小波变换，使得整个序列分解成不同频率的序列，同时利用小波分析在时域和频域上都具有良好的局部化性质，多尺度分析功能，结合ARMA模型的预测方法，以较为准确地根据历史数据预测其将来短期的走势。实验表明，MODWT时间序列分析方法比传统的时间序列分析方法预测的精度更高。

英文摘要：

To accurately predict the future of the stock price, a forecasting stock price method based on maximal overlap discrete wavelet transformation （MODWT） time series analysis is proposed （M-ARMA）. The mallat algorithm is used, and the MOD- WT are used in stock price series, so that the whole sequence are decomposed to different frequencies of sequence, meanwhile it have by using the wavelet it have not only the good localization properties in time domain and frequency domain, but also have multiresolution analysis function, combined with the method of forecasting by the ARMA model, it can accurately predict the fu- ture trend of short-term by the historical information. The test shows that the prediction accuracy of this method is higher than the method of traditional time series analysis.