To achieve efficient sharing and integrated development of financial information resources, the study proposes a blockchain-based financial management system. The system combines a server-side architecture with a modular data processing integration framework for seamless data flow, a dynamic game framework with group analysis strategy adaptation for dynamically mitigating 51% ledger attacks, and a hybrid tree structure for secure and efficient key-value storage in blockchain, as well as high fidelity data fusion for state trees, to solve the key challenges of secure and efficient financial data integration in enterprises. The results showed that the attack success rate was reduced by 91.2%, and the consensus latency was lower than 90 seconds (more than 10% higher than other consensus mechanism algorithms). The average data storage accuracy was over 90%, which was at least 5% more than that of cloud/distributed systems. The memory occupation result was no more than 1500 kb when the number of datasets was 5000, which was much smaller than 1906 kb and 2341 kb of distributed storage and cloud storage, respectively. The operation efficiency of the core financial function under the method was over 90%. The financial information management system can effectively add, modify, delete, and query voucher management, cashier management, and accounting information management. The proposed method can provide a secure and scalable solution for real-time financial decision-making in dynamic market conditions.