Forecasting the stock market is a difficult undertaking because of the intricate and ever-changing nature of financial markets. Long Short-Term Memory (LSTM) neural networks have displayed potential in grasping the time-related relationships in financial data for predicting prices. However, the effectiveness of LSTM models greatly depends on choosing the right hyperparameters. This paper introduces an innovative method for fine-tuning LSTM neural network hyperparameters by utilizing Taguchi Method to improve the accuracy of stock market predictions and additionally reduce the computational time required to obtain the best combination of hyperparameters. In this research, we establish a Taguchi method-driven optimization structure to automatically find the best combination of hyperparameters for LSTM models. Evaluations are performed using open-access datasets for stock markets with open prices.

Keywords