基于最优小波包变换、ARIMA与 SVR 的股票价格预测研究

摘要

股票价格序列的变化往往具有高度的非平稳性和异方差性，使得单一的预测方法难以准确预测。利用最优小波包变换，将股票价格序列分解为一系列特征规律较明显的小波包系数，对其中的趋势部分采用 ARIMA 进行预测，对细节部分采用 SVR 进行预测，最后将预测结果进行重构得到股价预测序列。实证研究结果表明：该预测方法结构明确，计算高效，能够以较高的精度对股价变化进行预测。%The changes in stock price series are always non -stationary and heteroscedastic,that fact makes single prediction method difficult to predict accurately.In this paper,using the best wavelet packet transform,the stock price series is decomposed into series of wavelet packet coefficients which reveal characteristics more obviously.The prediction of the trend coefficients using ARIMA Model,and the prediction of the detail coefficients using SVR Model,the predicted results are reconstructed to obtain the stock price forecasting sequence.The empirical result shows that,this mixed prediction method is of clear structure,high computation efficiency,and be able to predict the stock price changes with high accuracy.