Stock Trend Forecasting Using the ARIMA Model

Abstract

Stocks have always been a very important tool in the investment market. Nowadays, the stock market attracts a large number of investors as more and more people are exposed to investing their money. One of the most attractive features of equities for investors is the high returns, however, the high risks are also affecting investors’ confidence. Therefore, predicting the long-term performance of the stock market can lower the risk and give investors more ideas on how to invest and help them understand the future trend of their preferred stocks, thus reducing their risk rate. In this paper, the author proposed a stock predicting method based on the closing prices of Ford Motor Company over the past 50 years. In the selection of the model, the ADF test was utilized and by analyzing the dataset, the author demonstrated that the stock closing price is a non-stationary series and therefore the ARIMA model is selected. After differencing the series, a stationary series was obtained, and the best parameter was chosen based on Auto-ARIMA analyses. Furthermore, the author obtained the long-term trend graph based on the model output and analyzed the accuracy of the prediction results by RMSE and MAPE tests. In addition, by bringing data from other stock markets into the experiment, it is evident that the ARIMA model can be effective when predicting long-term trend of stocks and is able to be used as a method to forecast stock prices.