A Study on the Minimization of Derivatives Hedge Cost Using Artificial Neural Networks
Shin Hea Choi, Jae Pil Yu · Asia-pacific Journal of Convergent Research Interchange · 2024
This study empirically evaluated how effective machine learning is in the risk management of financial derivatives by comparing and analyzing traditional delta hedging strategies and machine learning-based hedging strategies for the KOSPI200 index option.The study's main purpose is to examine how machine learning models adapt to the complex structure and volatility of the market and how they can optimize hedge costs.To this end, the KOSPI200 index option data from November 2023 to September 2019 was used, and the hedge target value was predicted based on the artificial neural network (ANN).Dynamic hedging was performed based on this.As a result of the experiment, it was confirmed that the hedging strategy based on machine learning maintained a lower overall cost than delta hedging and superior adaptability to market volatility.These results show that machine learning has the potential to surpass traditional delta hedging methods, laying the foundation for further expanding the applicability of machine learning in financial risk management.This study also suggested that a machine learning hedging strategy can show stable and efficient performance under various market conditions.In addition, this study seeks directions for the continuous development and application of machine learning and proposes additional research in various asset groups and market environments.Future studies emphasize the need to evaluate the efficiency of risk management in a more complex market environment by applying more advanced machine learning techniques such as reinforcement learning.Such research is expected to contribute to establishing a new paradigm for financial risk management.