Forecasting Methods

The increased exposure of energy markets to free competition is a stylized fact for every fast-growing economy on the globe. Incremental competition undeniably increases the risk/uncertainty for every market participant (energy producers, energy utilities, etc.). In a risky environment, it is of an imperative importance for all market participants to make optimal decisions. Therefore, effective model building and precise forecasting for a target variable (energy prices, energy demand) have become a significant comparative advantage for extracting meaningful information from the available data. The course examines in-depth modern model building approaches and forecasting techniques, which are widely used in the energy sector. Topics which are examined include: exploring data patterns, moving averages and exponential smoothing, time-series decomposition, forecasting via dynamic regression models and cointegrated models. Additional topics include AR models, ARIMA models (Box-Jenkins method), univariate and multivariate GARCH models, state space models, non-linear models, long-term forecasting and forecasting evaluation. Provided that the orientation of the course is applied, emphasis will be given on the effective usage of powerful forecasting software and as a result several classes will take place on a lab. The skills developed in the course will be useful for managers and practitioners in maximizing the effectiveness of polices and business decisions. The student may apply into positions, which call for general data analysis and forecasting and monitoring market trends.