Stock Market Prediction Using Optimized Deep-ConvLSTM Model.

Area of Science:

• Financial Markets
• Artificial Intelligence
• Machine Learning

Background:

• Stock market prediction is complex, with many models failing to provide accurate forecasts.
• Accurate stock market prediction is crucial for investor profitability.

Purpose of the Study:

• To propose an effective stock market prediction system.
• To enhance prediction accuracy using a novel hybrid optimization algorithm and deep learning model.

Main Methods:

• A Deep Convolutional Long Short-Term Memory (Deep-ConvLSTM) model was employed for prediction.
• The model was trained using a hybrid Rider-based Monarch Butterfly Optimization (Rider-MBO) algorithm, integrating Rider Optimization Algorithm (ROA) and Monarch Butterfly Optimization (MBO).
• Technical indicators were computed, features were selected using Sparse-Fuzzy C-Means (Sparse-FCM), and robust features were fed into the Deep-ConvLSTM model.

Main Results:

• The proposed stock market prediction model achieved minimal Mean Squared Error (MSE) of 7.2487 and Root Mean Squared Error (RMSE) of 2.6923.
• Evaluation was performed on six live stock market datasets.

Conclusions:

• The proposed Rider-MBO optimized Deep-ConvLSTM model demonstrates significant effectiveness in stock market prediction.
• The system offers improved accuracy compared to existing methods, aiding investors in financial markets.