A method for identifying manipulative behavior in the electricity market based on the random forest algorithm considering both ex-ante and ex-post

The property rights of China’s power generation side are relatively concentrated, which makes it easier to engage in market manipulation. As the construction of the mid- and long-term electricity market gradually deepens, it is urgent to rely on key technical support to determine whether there is malicious market manipulation behavior. This article proposes a market manipulation behavior identification method based on the random forest algorithm that considers both ex-ante and expost. Nine indicators are used to extract features from the raw data, and a feature selection algorithm based on filtering and packaging is used to screen the sample features, thereby eliminating less important features in the sample. Finally, the model is trained based on the random forest algorithm to recognize market manipulation behavior, and the effectiveness of the proposed method is demonstrated through examples.