With the development of economy and the change of people investing consciousness, financial investment has become an important issue currently. Therefore, the financial prediction becomes an important investment tool to financial investors. Stock prediction plays a crucial role in a wide range of forecast in the financial market. It can also be extended to other fields of the financial forecast. In this paper, current stock forecasting methods are introduced first. Then a variety of prediction models are mainly introduced, which are the current popular four kinds of methods: BPN (back propagation network), ELMAN, SVM (support vector machine) and WNN (wavelet neural network). The cross validation method is added to find the optimal parameters in these four methods. Experiments with three different kinds of stocks are conducted to verify these four methods. The advantages and limitations of these methods are given by analyzing and comparing the experiment results.