Improving Stock Closing Price Prediction Using Recurrent Neural Network and Technical Indicators.

Area of Science:

• Computational Finance
• Machine Learning
• Time Series Analysis

Background:

• Accurate stock market prediction is challenging due to market volatility and external factors.
• Recurrent Neural Networks (RNNs) show potential for time series forecasting.
• Existing methods may not fully capture complex market dynamics.

Purpose of the Study:

• To propose a novel Long Short-Term Memory (LSTM) model for predicting stock closing prices.
• To enhance prediction accuracy by integrating technical indicators and basic trading data.
• To evaluate the model's performance on major stock market indices and a specific stock.

Main Methods:

• Utilized a Long Short-Term Memory (LSTM) network, a type of Recurrent Neural Network (RNN).
• Applied Principal Component Analysis (PCA) for dimension reduction of technical indicators.
• Employed optimization strategies including Adaptive Moment Estimation (Adam) and Glorot uniform initialization for model training.
• Conducted case studies on Standard & Poor's 500, NASDAQ, and Apple (AAPL).

Main Results:

• The proposed LSTM model demonstrated a good level of fitness in predicting stock closing prices.
• Comparison experiments using various evaluation criteria validated the model's performance.
• The methodology advanced the research in analyzing and predicting stock time series data.

Conclusions:

• The novel LSTM-based approach, incorporating PCA and Adam optimization, offers a robust method for stock market prediction.
• The model's ability to achieve a good fitness level suggests its potential for practical application in financial forecasting.
• This research contributes to the field of computational finance by providing an effective tool for stock time series analysis.