Comparison of Different Machine Learning Scenarios for Stock Price Prediction of Netflix

Abstract

Buying stocks based on relatively accurate predictions to make a profit is what investors have been yearning for. However, due to the volatility and stochastic intrinsic of the stock, the price is also full of uncertainty, which is difficult to predict. With the improvement of computer performance and the popularization of machine learning methods nowadays, effective stock prediction methods emerge one after another. In this paper, random forest, XGBoost, and LSTM techniques are utilized for predicting the closing price of Netflix, which is one of the major stocks in Nasdaq and has been fluctuating this year. The closing price of the previous five days and the last day’s volume are used as inputs. The models are evaluated by MSE and R-square. According to the analysis, similar model evaluation results in random forest and LSTM indicate that the models are efficient. Furthermore, the prediction could be improved by considering the fluctuations of stocks simultaneously and introducing more variables in various dimensions. These results shed light on guiding further exploration of predicting stock price movements based on advanced machine learning methods.