Affiliations:
Department of Mathematics, NED University of Engineering and Technology, Karachi, Pakistan
In the era of advanced computational techniques and predictive modeling, the focus has shifted toward reducing latency and minimizing costs. Financial market prediction is not a novel concept, as it has been applied effectively over the past decades to support informed decisions by traders and investors, leading to improved returns. However, machine learning and deep learning models often demand substantial computational power and processing time due to their complex architectures. Meta-learning provides an efficient alternative by reducing computation time and resource requirements for financial forecasting. This study proposes a meta-SGD model to predict future prices of the S&P 500 index and NASDAQ, and compares its performance with deep learning models (CNN and GRU) and a hybrid CNN-GRU model. Evaluation using RMSE, MAE, and R² metrics shows that the meta-learning model outperforms both deep learning and hybrid models, achieving state-of-the-art predictive accuracy with significantly lower computational cost.
Financial market forecasting, Meta-learning, Deep learning, Computational efficiency, Predictive modeling
https://doi.org/10.21833/ijaas.2025.11.008
Batool, K., Baig, M. M., & Fatima, U. (2025). Meta-learning for financial market prediction: An efficient approach with reduced computational cost. International Journal of Advanced and Applied Sciences, 12(11), 72–81. https://doi.org/10.21833/ijaas.2025.11.008