Security investment prediction based on wavelet analysis theory

Applying wavelet analysis, a method is proposed to model system identification for stock market investment prediction. The suggested method includes two steps. First, the data is filtered by wavelet decomposition to eliminate the high frequency noise and to improve the precision. Second, the mathematical model prediction is carried out by Least Square (LS) for price change of medium or long-term security. The efficiency of the method is proved by simulation. This suggests that the method is simple and gives prediction values better than those by the LS method.