The comprehensive energy resource management presents kind of global optimisation, which is much more cost-effective than separate treatment of energy-demand forecasting and energy-inventory data. In this paper, two-stage supply chain is presented, where distributor of derivate energy products (e.g. hot water, steam, electricity) shares information with its supplier of primary energy sources (e.g. coal, crude oil, gas). Different methods of energy-demand forecasting can be used, although exponential smoothing methods are most often used in practice because they are simple, fast and inexpensive. In this study we analysed a modified Holt-Winters (HW) method and we describe a method for simultaneous optimisation of forecasting method and a stock control policy. We compared the proposed joint optimisation of total cost with the optimisation, based solely on forecasting data. The total cost presents the joint cost of distributor and supplier. The data consisted of 756 quarterly real series from M3-Competition and 300 simulated demand patterns. We have shown that the total cost can be reduced dramatically if we use the joint optimisation instead of separate treatment of forecasting method and inventory model. We obtained the best result with the modified HW method; the essential reduction of total cost was reached in case of simulated data as well as in case of real data.