Forecasting Natural Gas Prices with Spatio-Temporal Copula-based Time Series Models

TL;DR

This paper introduces spatio-temporal copula-based models for forecasting complex commodity prices, addressing heavy tails, skewness, and non-linear dependencies, and enhances point forecasting with neural networks.

Contribution

It presents a novel application of copula models to commodity price forecasting and integrates neural networks for optimal point forecast determination.

Findings

Copula models are competitive in forecasting fossil fuel and emission prices.

The use of neural networks improves point forecast accuracy.

Non-elliptical probabilistic forecasts are effectively visualized.

Abstract

Commodity price time series possess interesting features, such as heavy-tailedness, skewness, heteroskedasticity, and non-linear dependence structures. These features pose challenges for modeling and forecasting. In this work, we explore how spatio-temporal copula-based time series models can be effectively employed for these purposes. We focus on price series for fossil fuels and carbon emissions. Further, we illustrate how the t-copula may be used in conditional heteroskedasticity modeling. The possible emergence of non-elliptical probabilistic forecasts in this context is examined and visualized. The problem of finding an appropriate point forecast given a non-elliptical probabilistic forecast is discussed. We propose a solution where the forecast is augmented with an artificial neural network (ANN). The ANN predicts the best (in MSE sense) quantile to use as point forecast. In a forecasting study, we find that the copula-based models are competitive.