Drilling Deeper: Non-Linear, Non-Parametric Natural Gas Price and Volatility Forecasting

• DOI: 10.5547/01956574.45.4.dbaj
• Publication Date: 2023-10-25T14:12:07Z
• Authors: Dusan Bajatovic; Deniz Erdemlioglu; Nikola Gradojevic
• Keywords: Natural gas prices

This paper studies the forecast accuracy and explainability of a battery dayahead (Henry Hub Title Transfer Facility (TTF)) natural gas price volatility models. The results demonstrate dominance non-linear, non-parametric models with deep structure relative to various competing model specifications. By employing explainable artificial intelligence (XAI) approach, we document that is formed strategically based on crude oil electricity prices. While conditional returns driven by long-memory dynamics volatility, informativeness predictor has improved over most recent volatile time period. Although reveal predictive non-linear relationships are inherently complex time-varying, our findings in general support notion gas, interconnected. Focusing periods when markets experienced sharp structural breaks extreme (e.g., COVID-19 pandemic Russia-Ukraine conflict), show learning provide better adaptability lead significantly more accurate performance.