Transfer Learning for Volatility Pattern Prediction Across Global Stock Markets Using Lightweight Statistical Models

Transfer Learning for Volatility Pattern Prediction Across Global Stock Markets

Transfer Learning for Volatility Pattern Prediction Across Global Stock Markets Using Lightweight Statistical Models

International Journal of Computer Techniques – Volume 12 Issue 3, May – June 2025

Authors

R. Harini – Research Scholar, Dept. of Computer Applications, Shrimati Indira Gandhi College, Tiruchirappalli, India. harinidr.research@gmail.com

Dr. S. Rethinavalli – Research Supervisor, Dept. of Computer Applications, Shrimati Indira Gandhi College, Tiruchirappalli, India. rethinasowri@gmail.com

Abstract

This research proposes a **statistical transfer learning framework** using **ARIMA and Exponential Smoothing models** to predict volatility across global stock markets. The methodology aims for **cross-market adaptability** while maintaining **low computational overhead** and **high interpretability**. Experimental validation on **NYSE, BSE, and FTSE datasets** shows strong performance, suggesting a scalable alternative to deep learning in data-constrained financial environments.

Keywords

Transfer Learning, Volatility Prediction, ARIMA, Exponential Smoothing, Global Stock Markets, Financial Time Series, Lightweight Models, Interpretability.

Conclusion and Future Work

This study offers a novel **statistical transfer learning system** that balances predictive accuracy and transparency across diverse financial domains. Future work may involve the **integration of macroeconomic features**, adaptive windowing, and **ensemble modeling** to extend use cases to emerging markets and highly volatile sectors. A **cloud-based decision-support platform** will operationalize this framework for real-world fintech applications.

References

  1. Zhang, Y. & Zhou, D. (2020). Transfer learning for cross-market stock prediction. ESWA, 139. https://doi.org/10.1016/j.eswa.2019.112852
  2. Patel, J. et al. (2015). Predicting stock market index using fusion of ML techniques. ESWA, 42(4), 2162–2172.
  3. Chen, Y. et al. (2021). Transfer learning for financial time series forecasting: A survey. arXiv:2105.06895
  4. Kumar, S. & Bhattacharya, S. (2022). Lightweight statistical models in financial forecasting. Journal of Forecasting, 41(5), 800–814.

Post Comment