“…Intra-day power prices have a fine structure and are driven by an interaction of fundamental, behavioural and stochastic factors. Moreover, the dynamic analysis of model specifications appears to suggest both evolutionary and abrupt changes in the impact of exogenous drivers (Chen and Bunn, 2010;Gonzalez et al, 2012) and, in particular, substantial changes in fuel and carbon allowance prices as well as water flows, solar and wind production can, for example, change the profile of the price-setting power plants.Research on the dynamics of intra-day price formation is therefore strongly persuasive of the benefits of nonlinear models and, while the dominant suggestions are towards the general class of regime-switching methods, it remains as an open question, whether the apparently better in-sample fits and explanatory power, ex post, of nonlinear, multi factor methods, translates into more accurate, robust ex ante price forecasting (contrast the positive evidence for regime switching in Gonzalez et al, 2005Gonzalez et al, , 2012Bierbrauer et al, 2007;Bunn, 2008b, andChen andBunn, 2010, with the results in favour of forecasting with linear models in Haldrup and Nielsen, 2006;Kosater andMosler, 2006), andMisiorek et al, 2006). Markov regime switching has been widely advocated to capture some aspects of the nonlinearity, but it may suffer from overfitting and unobservability in the underlying states.…”