Abstract
Analyzing the U.S. stock market is a big data challenge due to its high volume of data and large price volatility. This project demonstrated how deep learning, specifically the Long Short Term Memory (LSTM) recurrent neural network, could make use of the huge amount of financial data to predict stock price movement. The main goal was to help investors make better stock trading decisions and increase their stock portfolio’s return by forecasting future stock price trends in 1-month periods based on historical market data. The applied learning model used stocks selected from various market sectors as inputs and forecasted the individual stock price trend for each sector separately. To evaluate the model, experimental results were compared to determine which market sectors were most effectively predicted, based on these three metrics: predicted against actual plot, accuracy of price movement, and root mean squared error. The experimental results showed that the future stock price trend in certain market sectors are more effectively predicted than in the other sectors.