Decision Trees auf der Flame 2019

Decision Support for Asset Backed Gas Trading based on Stochastic Optimization

Managing gas storages has become to a difficult task in recent years. Very uncertain medium to long term movements of gas market prices make hedging of gas storage positions very risky, whereby at the same time revenues that cover the storage's costs are low. In particular, riskless strategies that apply rolling intrinsic back-to-back trading, are not yielding revenues that cover the costs for the storage.

Decision Trees GmbH is presenting an innovative stochastic model for gas storage and gas contract hedging and operation at this year's Flame conference in Amsterdam. This model has been developed in close collaboration with a major world-wide operating oil and gas enterprise. It is meanwhile in productive use at various gas asset operating utilities in Germany, Austria and the UK. The outperformance that has been realized over traditional rolling intrinsic approaches is significant. 

The model is based on trinomial scenario trees which cover not only future day ahead price scenarios but scenarios of the whole forward curve day by day. The model holds decision variables for both, physical gas storage operation and managing hedging positions. Therefore, decisions for injection/withdrawal and buying/selling different products in the market are proposed by one single stochastic model.