Journal of Energy Markets

Derek W. Bunn
London Business School

This issue of the journal contains four strong papers applying advanced econometric, time series and financial engineering techniques to the main energy markets for electricity, oil and gas.

In the issue’s first paper, “A market scoring mechanism for trading of German electricity futures” by Tarjei Kristiansen, a novel systematic commodity trading model is developed utilizing a time series momentum strategy. The paper’s main innovation is a scoring mechanism to generate buy and sell signals, as well as determining the position sizes. The model is applied to the German electricity futures market and uses the following price drivers: momentum and volatility of the German front quarter, Argus/McCloskey’s Coal Price Index (API 2) and emissions. Tested on data from November 2010 to December 2019, the model outperforms a basic short sale model, delivering improved risk-adjusted returns.

In “Using equity, index and commodity options to obtain forward-looking measures of equity and commodity betas and idiosyncratic variance”, another paper related to the forward markets, Ehud I. Ronn presents a parsimonious and theoretically sound basis for extracting forward-looking measures of betas and idiosyncratic variances. Defining forward-looking betas and idiosyncratic variances as perturbations of historical estimates, Ronn applies equity and index options under single-factor models to discern market perceptions of oil companies’ prospective betas. The prospective fraction of idiosyncratic variance relative to total variance indicates the onset of crises when the idiosyncratic component fades relative to the systematic effects. This complements Chicago Board Options Exchange (CBOE) Volatility Index (VIX) and CBOE implied correlations. The model is extended to options on equities and oil futures. This forward-looking oil beta, in conjunction with risk-neutral futures prices, yields a capital asset pricing model based oil price forecast.

Moving from forward prices to trading, Cliff Parsons develops some theoretical concepts for optimal injecting and withdrawing activities for gas storage in his paper “Theory for optimizing capacitated commodity storage with case studies in natural gas”. He starts with the concept of spot price thresholds. These represent the expected value of future opportunities forgone or created by injecting or withdrawing any commodity currently. Comparing the spot price with the spot price thresholds dictates when current trading is optimal. Under certain conditions, these thresholds are monotonically decreasing in inventory level. As such, they relate to and extend previous work that proves the existence of two critical inventory levels such that injecting, withdrawing and doing nothing, respectively, is optimal for current inventories less than, greater than and between those levels. This paper shows that the current spot price in relation to these thresholds both causes these results and can be used to calculate the critical inventory levels. Also, the gap between the two critical inventory levels is driven mainly by transactions costs.

When it comes to forecasting gas prices, the related paper by Francisco Castañeda, Markus Schicks, Sascha Niro and Niklas Hartmann, “Forecasting natural gas price trends using random forest and support vector machine classifiers”, provides a modern data science approach to a well-established problem. In this work, different machine learning approaches are applied to forecast future yearly price trends in the natural gas Title Transfer Facility market in the Netherlands. The study compares two models: random forest and support vector machine classifiers. Castañeda et al find that the identification of potential gas price drivers is a crucial aspect of relative accuracy. Machine learning is becoming more widely used for forecasting and to support algorithmic trading in energy markets, but the techniques need constant updating to remain state of the art.