Forecasting the Stock Market Index with Dynamic ARIMA Model and LSTM Model

With the development of the machine learning method, there are a lot more time series model being invented and applied to mimic the real-world data. The interpretation and prediction of time series in financial markets is a hot topic in current research. This thesis conducts dynamic ARIMA model and the Long-short term model to forecast the stock market index in America and check the causal inference between the residual of the forecasting and the federal fund rate, which could explain the abnormal increase in the period 2021–2022. Thus, this paper provides a hybrid explanation of the structure of the time series forecasting, which will be helpful with the predicting. And this thesis also shows that the epoch for long-short term need to be considered when concluding in a common result of forecast. The deep learning method should be more accurate with a vast data set and become more helpful. This study provides a new idea for the prediction of the US stock market index through the comparison of prediction results between models, expanding the current research field.