As quantitative consultants and model developers of the UK, Continental, and North American commodity markets, we often get the question whether or not the values from storage valuation models, including ours, should be trusted. In particular, the extrinsic spot-based value is often considered as a mystical number that is too good to be true. In order to find a compelling answer, we therefore performed a large-scale backtesting study of the UK gas market, the most liquid in Europe. In this article, we share the results and our insights.

The backtest looks at the performance of different storage trading strategies in the UK NBP gas market. We assess a storage bundle mimicking the characteristics of the Rough storage, the largest in the UK. The backtest period covers 12 years, from 1997 until 2008. Every half year, the storage model calculates the expected storage value over the forthcoming 12 months. This 'projected' present value (ppv) is either the intrinsic value (based on the current forward curve), the rolling intrinsic value (based on changes in the forward prices over time), or the spot-based value. In the backtest, we then carry out the underlying trading strategies in the market over the front 12 months.

Our results indicate that the profitability of storage trading has varied largely over time, mainly due to variations in winter-summer spreads and price volatilities. When a trader would have relied on a pure spot trading strategy only, he would have done very well in some years, but in fact, often performed below expectation (the ppv being the expectation). This research discusses various explanations. However, a combination of a spot trading and a forward market hedging strategy completely changes the picture. The trader performance then matches closely with the expectation and he can be quite sure to realize both the intrinsic and the extrinsic value.