Short-term modeling of carbon price based on fuel and energy determinants in EU ETS

Accurate price forecasts on the European Union Emissions Trading System (EU ETS) market are of interest to many production and investment entities. The article presents a new approach, which models the most significant determinants by the use of a rule-based model containing facts written in the form of ordered fuzzy numbers, as the description of trends of imprecise values of prices. This approach allows for a qualitative analysis of potential values and directions of EU allowance (EUA) price changes and their trends along with the probability of their occurrence in the future. Article presents a wide set of factors used for the analysis of EUA price conditions and the comparison of the results of various feature selection methods (Hellwig's method, F-test, Neighborhood Component Feature Selection method, RreliefF method). Another scientific value added to the presented research results is a new approach to modeling an influence of the significant determinants for a day-ahead prediction CO2 price in the form of fuzzy rules which contain additional information about the tendencies of value changes. Thanks to a comprehensive approach, improvements to existing methods, and proprietary models, additional information was obtained about the tendencies of EUA price level changes.