A Hybrid Quantum-Classical Model for Stock Price Prediction Using Quantum-Enhanced Long Short-Term Memory

Area of Science:

• Quantum Computing
• Machine Learning
• Financial Markets

Background:

• Stock market prediction is a complex challenge in machine learning (ML).
• Classical ML models for prediction are computationally intensive.
• Quantum computing (QC) offers potential for exponential speedups over classical computers.

Purpose of the Study:

• To develop and evaluate a hybrid quantum-classical ML model for stock price prediction.
• To introduce a novel model, Quantum Long Short-Term Memory (QLSTM), by integrating classical Long Short-Term Memory (LSTM) with QC.
• To compare QLSTM's performance against classical ML models.

Main Methods:

• Developed a hybrid quantum-classical ML model (QLSTM).
• Validated QLSTM using an IBM quantum simulator and a real IBM quantum computer.
• Evaluated performance using Root Mean Square Error (RMSE) and prediction accuracy.
• Conducted comparative analysis against classical models and explored hyperparameter impact.

Main Results:

• QLSTM achieved a lower RMSE (0.0602) compared to classical LSTM (0.0693).
• QLSTM demonstrated higher prediction accuracy (0.9736) than classical LSTM (0.8815).
• QLSTM outperformed other classical models in both RMSE and accuracy metrics.

Conclusions:

• The hybrid QLSTM model shows superior performance for stock price prediction.
• Integrating QC with classical ML offers significant advantages in financial forecasting.
• QLSTM represents a promising advancement in applying quantum computing to financial markets.