Forecasting electricity prices and market length for trading stochastic generation in markets with a single-price balancing mechanism

This paper presents the problem facing stochastic generators who are required to participate in electricity markets with a single-price balancing mechanism. The solution to this problem requires forecasts of multiple processes: weatherdependent generation, electricity and balancing prices, and the sign of the system length. By formulating the problem from a probabilistic perspective, it is demonstrated that a combination of well known and understood forecasting techniques can support market participants in both increasing revenue and reducing risk. Probabilistic forecasts of system length are produced using logistic regression on data widely available to market participants, and electricity prices are forecast using ARMAX models with automated fitting. Wind power forecasts are provided by a wind farm operator for a case study based on wind participating in the UK electricity market. It is shown that the proposed approach can be employed to increase revenue, by over 10% in the most extreme case, and to reduce risk