Financial derivatives were born for the purpose of reallocating risk resources, often through the purchase of financial derivatives, a certain cost of funds in exchange for the transfer of risk. 3214 enterprises are selected for data preprocessing, 113 indicators of credit risk of financial derivative products are selected for the first time, including profitability, solvency, and macroeconomic environment and other aspects, and principal component analysis is used to determine 8 effective assessment indicators. The XGBoost algorithm is introduced, and the credit risk prediction model of XGBoost-GP is constructed by combining the Bayesian optimization method based on Gaussian process, and by setting up the comparison model, the XGBoost-GP obtains the best prediction accuracy of 84.7% on the dataset, and the XGBoostt-GP model has a fast convergence speed. By combining XGBoost-GP and SHAP method to establish a credit risk assessment model, the model is utilized to evaluate the indicators from the perspective of evaluation indicators, the model can ensure the high accuracy and robustness of the assessment results, profitability plays an important role in the assessment of credit risk of financial derivatives, and the assessment system provides a reference method for the control of risks in the financial derivatives market in the era of digital economy.
