Stock Price Research Based On ARIMA Model

With the rapid development of science and technology, China’s stock market has gradually become the main capital market in our country. Whether stock investors or managers, they are looking forward to getting a great reward with small investment. As information is changing all the time, it will lead to the price of stock changes too. Through analyzing the stock price, stock market information can be found, and through time series analysis, some useful information can be mined. Thus, it can provide reference for stock investors and managers and help them achieve the maximum return.First, it’s be found that in some articles data are used to build models and before that there is no test on data if they are appropriate for the models, such as reference[7] and [8]. This paper has analyzed the data of reference [7]、[9]、[11]、[12] and the opening price of China Merchants Bank and PING AN Bank of a certain period of time, also combined with the white noise test, it’s found that the eight sets of data does not fit for ARIMA model. By understanding the theory and modeling of the GARCH model idea, based on GARCH models, this paper analyze and research on these data, and it’s confirmed that five sets of data are fit for GARCH model. And, using exponential smooth method and SAS software to realize, it’s found that the remaining three sets of data can be built the exponential smoothing model.This paper established models on the basis of testing on data, it’s confirmed that the data are really fit for the model. Test on data should be the basis and prerequisite for modeling. When analyzing the data, we should build models according to the characteristic of the data itself, and check if the data is suitable for building such a model or not. On this basis, the model’s fitting and accuracy can be improved, and the built models will be more feasible.