Stock market is complicated dynamic system affected by many factors. To model it, many methods have been proposed. But those methods cannot solve this problem very well. In this paper, from analyses the mathematic description of stock market system, a new method based on new evolutionary neural network is proposed here. In this new evolutionary neural network, the traditional BP algorithm and a new bionics algorithm, immunized evolutionary programming proposed by author 1 is combined. In order to verify this new method, the stock market data of Shanghai market in 1996 is used. The results show...