Data forecast is important in the field of economic investment, and the price of the stock is one of the most the complex data types in economic data. This paper uses cultural algorithms to establish time series models, focuses on the characteristics of the stock data to select a cultural algorihtm model which is fit for the stock rules,and  comparies with the result which is obtained by linear regression algorithms, in order to analyse the merits of the application of cultural algorithms for the problem . The result shows that using cultural algorithms models for forecasting achieves better results, and the forecast accuracy is high.The relative error of the forecast is within 3%, and the rate of correct analysis is more than 70%.