Analysis of demand and supply curves in the Italian natural gas market through functional time series

Motivated by price prediction in Italian natural gas balancing market, we propose a model to forecast supply and demand curves evolving through time. Differently from the previous literature, where supply and demand curves are described as static objects, here we consider them as dynamic objects. Assuming that the future curve depends on the previous curves and on some quantities available at prediction time, we looked for a suitable functional autoregressive model with scalar covariates which maintains the monotonicity constraints. The results are satisfactory and coherent with standard time series analysis but with much richer information about the market dynamic.