Currently, China′s electricity market is in a phase of deepening reforms, where collusion between the generation side and the retail side intensifies the risk of market power abuse, severely hindering the market-oriented development of the power sector. To enhance the prevention of collusion between the generation and retail sides in the electricity marketization process, a progressive bilateral matching model for market power risk identification is proposed. This model is applicable to medium- and long-term trading scenarios in the electricity market and aims to identify market power risks among power generation companies and retail suppliers. By integrating the analysis of subjective and objective behavioral characteristics of market entities, the method establishes a comprehensive identification framework of "risk feature screening - abnormal transaction locking." It employs the Borda counting method to quantify subjective preference features of the generation and retail sides across three dimensions—pricing, volume bidding, and cooperative stability—while introducing an entropy-balanced matching algorithm to objectively quantify the concentration of clearing results. This ultimately enables precise identification of at-risk entities. Case study results demonstrate that the method not only accurately identifies high-risk trading entities but also effectively enhances market supervision efficiency.