A functional time series is a sequence of observations in which each observation is a function that takes values in a closed interval. These time series appear frequently in electricity markets, e.g. the electricity demand, the electricity price, offer curves or Residual Demand Curves. Being able to forecast these series is of utmost importance for market agents.

This PhD thesis proposes a novel method for forecasting functional time series. This model is designed for modeling the complex features of functional time series in electricity markets such as seasonality and dependencies on explanatory variables.