Financial transaction transparency has gradually become one of the main directions for the development and construction of the financial transaction market. This paper integrates blockchain and smart contracts and proposes a strategy to improve financial transaction transparency in order to protect transaction data privacy and identify and trace transaction anomalies. The proposed DM-IBBE scheme for smart contract transaction privacy involves choosing different interpolation points based on Lagrange interpolation curves and creating encryption modes that meet the requirements for financial transaction privacy. Based on a graph neural network, the propagation probability of abnormal transactions is calculated from the blockchain network topology using the TAGCN model, and the influence of irrelevant noise pairs is eliminated to realize the identification and traceability of abnormal transactions. Taking the financial transaction platform of City A as the research object and carrying out the practice of financial transaction optimization, the evaluation scores of the first-level indexes of comprehensive government transparency, transaction process transparency, operation result transparency, process service transparency, and operation transparency and guarantee are 76.58, 88.93, 95.42, 89.51, and 88.43, and except for the indexes of comprehensive government transparency, the other indexes are all greater than 80 points. The second-level indicators’ evaluation value increases from the 50–70 score range before financial transactions optimization to the 80–100 score range. The financial transaction platform in City A has significantly improved the transparency of financial transactions.