DATS6501-Capstone Project-Zixuan Huang

In this research, the objective was to utilized LSTM based models for stock trend predictions. For my study, I have used S&P 500 index values for 20 years as data. An RNN model was used as a benchmark. In comparison, four kinds of LSTM models, including Vanilla LSTM, Stacked LSTM, Bidirectional LSTM and CNN-LSTM were built. In this work, the input was the daily close price of the S&P 500 index, the output of the model was compared with the actual stock price data by using root mean squared error.