This article selects 9 real estate companies that were roughly the same in ST time in 2021 and another 31 real estate companies with good financial conditions. Select the financial indicators from T‐3 to T‐1 as the sample data (the time for selecting the financial indicators of non‐ST companies is the same). Use SPSS 26.0 to analyze the significance of the selected financial indicators, and then use principal component analysis and Logistic regression to establish two types of models: one, financial variable early warning model, second, comprehensive early warning model. The research conclusions show that the accuracy of the early warning model is improved after the addition of non‐financial indicators. The introduction of non‐financial indicators can increase the ability of the financial early warning model to discriminate future conditions.