Abstract: This paper proposes the forecasting method based on wavelet transform and MOBP.By using the function of wavelet transform，the non-stationary time series of the stoke price can be decomposed into approximate series and several detail series，then build the forecasting model on MOBP，carry out the training of MOBP.The prediction of the original series can be obtained by the synthesis of prediction result of each series.The experiment results show that the forecasting performance on this method is good，and can obtain the higher forecastin precision than the conventional BP neural network.