Abstract: This paper proposes a financing risk evaluation system following the idea that the financing risks should be divided,apportioned and evaluated. Firstly,the financing risks are divided into risks that might emerge in negotiation phase and performance phase following New Two-stage Theory. Then a risk apportionment model combing Grey Relational Degree and TOPSIS is constructed. Afterwards,a financing risk evaluation model is built by integrating Genetic Algorithm and Support Vector Machine (SVM). Finally,Qingdao Jiaozhou Bay Subsea Tunnel is exemplified in the empirical study,and the final results indicate that the evaluation system is reliable,which supports the individual stakeholder of large-scale infrastructure projects to grasp the risk situation in time.