A Combined Stock Price Forecasting Model Based On The Wavelet Packet Denoising

The purpose of this article is to use the combined model which based on the wavelet packet denoising combining with BP neural network model, ARMA model and exponential smoothing (ES) model to examine the effectiveness of forecasting s-tock price. The selection datas are the three years’daily closing price of the China construction bank’s stock price in2011-2013. First, with the original data of con-struction bank, three single models, BP model, ARMA model and ES model are set up respectively. Then particle swarm optimization (PSO) algorithm is used to optimize the weights of the combined model. And we construct the combined model and predict for datas by making use of the three single models and the combined model. It finds that the effectiveness in predicting the stock price of the combined model is better than any other single model. Then with the datas after wavelet packet denoising of China construction bank, the same three single models and the combined model are set up, and predict for the obtained datas. According to the results, based on the wavelet pack-et denoising model’s prediction effectiveness is more better than the unwavelet packet denoising model. Thereby the combined model based on wavelet packet denoising is effective for stock price forecasting.