The implementation of smart algorithms to support financial planning is increasingly widespread in corporate activities. As presented in this paper, the Metric-Based Decision Tree (MBDT) algorithm is introduced to measure the information gain rate of financial data on solvency, operational efficiency, profitability, and growth potential. This process enables data attribute segmentation and node classification. Based on a backtracking corporate financial decision tree, the algorithm achieves financial crisis early warning and decision support. The MBDT algorithm achieves a maximum financial data classification accuracy of 92.125%, with a goodness-of-fit error not exceeding 5.815%. In practical applications, it can analyze and forecast a company’s financial data spanning the past 10 years and the next 10 years. Setting the critical threshold to 0.60 yields the highest precision in financial crisis early warning.
