A Long Short-Term Memory Network Stock Price Prediction with Leading Indicators.

Area of Science:

• Financial forecasting
• Computational finance
• Machine learning applications in economics

Background:

• Accurate stock price prediction is vital for investment strategies and fund management.
• Traditional forecasting methods struggle with the complexity of financial markets and numerous influencing factors.
• The integration of leading indicators is explored to enhance predictive accuracy.

Purpose of the Study:

• To develop and evaluate a novel stock price prediction framework, LSTMLI, integrating leading economic indicators.
• To assess the impact of leading indicators on stock market volatility prediction.
• To compare the performance of the LSTMLI model against other neural network models.

Main Methods:

• Development of a Long Short-Term Memory (LSTM) based framework named LSTMLI.
• Inclusion of leading economic indicators alongside historical stock data, futures, and options.
• Application of a classification method for qualitative stock price movement prediction, instead of regression.

Main Results:

• The LSTMLI model, utilizing a classification approach, demonstrated a significant reduction in prediction errors.
• The inclusion of leading indicators in the dataset improved LSTMLI's predictive performance compared to using historical data alone.
• LSTMLI showed competitive predictive performance relative to other neural network models.

Conclusions:

• The LSTMLI framework effectively predicts stock price fluctuations by incorporating leading economic indicators.
• The classification method enhances the accuracy of stock trend prediction, offering qualitative insights.
• This approach provides a more robust tool for investors and financial researchers in dynamic markets.