Philippine Stock Exchange Index Forecasting Using a Tuned Artificial Neural Network Model with a Modified Firefly Algorithm

Forecasting stock market index fluctuations is crucial for financial market participants. However, analyzing time-series datasets from stock markets is challenging due to their noisy and dynamic nature. Artificial neural network (ANN) models have shown promising results in time-series forecasting, but their success heavily relies on carefully selecting and configuring their hyperparameters. This process often involves time-consuming trial-and-error techniques, default parameter usage, or reliance on limited existing studies that may have biases and errors. To address this, this research proposes a novel approach using a modified Firefly algorithm to automatically find optimal hyperparameters for the ANN model. The method is applied to forecast the Philippine Stock Exchange Composite Index (PSEi) closing price for the next day using data from May 11, 2018, to May 10, 2023. Results show minimal error rates, confirming the approach's effectiveness in improving forecasting techniques and enhancing decision-making in the ever-changing stock market landscape.