Predicting High-Frequency Stock Market Trends with LSTM Networks and Technical Indicators

The application of deep learning in stock price prediction has actually obtained boosting importance because of its capacity to capture complex patterns in financial information. This study uses Long Short-Term Memory (LSTM) networks to anticipate stock costs utilizing both day-to-day trading information and high-frequency minute-level information. For the day-to-day trading data, 42 stock datasets were analyzed, with 37 utilized for training and 5 booked for testing. High-frequency data, characterized by minute-by-minute changes, called for the development of specific designs for each stock. Each design incorporated differing configurations of LSTM devices and technical indicators to enhance predictive accuracy. The experimental outcomes demonstrated that the inclusion of technical indicators led to a reduction in prediction error, as measured by Root Mean Squared Error (RMSE). This highlights the capacity of LSTM networks, especially when boosted with technical indicators, for improving stock price forecasting accuracy throughout various timeframes.