Forecasting day ahead spot price movements of natural gas – An analysis of potential influence factors on basis of a NARX neural network
Using a dynamic forecasting approach on basis of a NARX neural network, the possibility of forecasting the day ahead spot price movement of natural gas in the market area of NetConnect Germany is examined. For this purpose, several trader interviews brought together potential influence factors such as temperature, cross-border flows or storage withdrawal rates. To determine the optimal variable combination, those influence factors are used in a sensitivity analysis to estimate the individual impact. It could be shown, that a NARX net depending on only five factors (temperature, exchange rate and the settlements of three major gas hubs) produces the most accurate forecast. As the dominating influence factor the temperature forecast four days ahead (t+4) could be identified.
Use and reproduction:
All rights reserved