Stock Price Prediction using Deep Neural Network/LSTM* Machine Learning Model

*Multivariate Time Series Forecasting with Long-Short Term Memory (LSTM), the variant of Recurrent Neural Network (RNN): stock prices and trading volumes data ('Open', 'High', 'Low', 'Close', 'Volume') for previous 90 business days are used to predict adjusted closing stock price ('Adj Close') for next 3, 5, or 10 days. The model is trained over the up-to-date stock data for the past 5 years every evening (around 6:30 PM EST or 11:30 PM UTC/GMT).

Note that the machine learning model results are shown for the demonstration purpose only, not for financial advice. Stock price prediction is a very challenging task since stock market depends on multiple factors that tend to be volatile and unpredictable. The machine learning model used in this work is relatively simple and limited, due to the limited resources (e.g., stock market-related data, computational resources, etc.). In turn, the model results in relatively high inaccuracies especially for the prediction of many days in the future.