Stock Price Prediction Based on Time Series Model

Due to the nature of the stock market, we know that stocks fluctuate violently on a daily basis, not only affected by the market environment, but also by some political factors and natural disasters, therefore it is difficult to predict the rise and fall of stocks. I have been looking for a more accurate way to predict the changes of the stock market. In order to achieve the goal, I choose the optimal stock prediction method by comparing different prediction methods.In recent years, the quantitative investment has developed rapidly. Due to the stable performance of the quantitative investment trading strategy, its market scale and share are constantly expanding, which has been considered by more and more international investors.In the following article, I will use advanced models and machine learning algorithms to predict the rise and fall and the trend of MSFT stock price in the future which based on the data of MSFT stock price obtained in previous years. I mainly selected three models for comparison. GARCH family model, Auto ARIMA model and LSTM neural network model.I first process the data, remove the missing values and segment them. For GARCH, I use Eviews to model. When building the Auto ARIMA and LSTM model, I read the data into Python and segment the data. Next, I set the neuron points, iteratively train the data, and predict the trained data to fit the image. Finally, after comparison, I choose the LSTM model with better fitting degree to make prediction.