Review of AI and Machine Learning-Based Hedging Strategies in Commodities Derivatives

Abstract

This study reviews the paradigm of AI and Machine Learning (ML)-based models in hedging strategies in the Commodities Derivatives Market with aim to describes the research conducted so far, methodological approaches, and a new model that is empirically explored. For this purpose, PRISMA Methodology is followed in data extraction, and a rigorous content analysis-based systematic literature review is done. A total of 20 peer-reviewed empirical articles were analysed from the Scopus and Web of Science databases. The review shows that USA’s exchange lead in the application of AI, Deep Learning and ML-based models and crude oil (including WTI and Brent) - asset class is used most frequently. However, fewer studies are available to support the empirical validity of these models in developing economies. While DL and ML models like “GJR-GARCH-SVR-LSTM”, “Deep Neural Networks” and LSTM”, “VaR with Random Forest model” and other machine learning-enhanced VaR models, are utilized in financial hedging strategies, but their application and empirical testing in commodity derivatives remain unexplored. The findings indicate the need to include agricultural products and industrial metals as units of analysis in these ML and DL models. The results of this review may serve as a valuable resource for researchers and managers seeking to conduct further investigations into this head of study.