The Evolution and Forecasting of Sino-Korean Commodity Trade: A Theoretical Exploration Based on Garch and Machine Learning Models

This paper examines the evolution of Sino-Korean commodity trade and explores the application of advanced econometric and machine learning models for forecasting commodity price volatility. With increasing interdependence between China and South Korea in trading key commodities such as crude oil, LNG, iron ore, and rare earth elements, accurate price forecasting has become crucial for managing economic risks and optimizing trade strategies. The study highlights the limitations of traditional econometric models like GARCH, which, while effective at capturing short-term volatility, struggle to account for the complex, nonlinear dynamics present in modern commodity markets. Machine learning models, including LSTM, random forests, and support vector machines, offer a more flexible and accurate approach by incorporating real-time data and adapting to market shifts. The combination of GARCH and machine learning in hybrid models further enhances forecasting accuracy. As both countries transition toward sustainable energy, the role of advanced forecasting tools will be pivotal in maintaining economic stability and fostering deeper trade cooperation.