Regression models in the context of digital economy play an important role in financial risk assessment. This paper introduces the DY spillover method in TVP-VAR and establishes the time-varying parameter vector autoregressive spillover index model (TVP-VAR-DY) to measure the level of risk spillover. Based on this, the model evaluation index system is constructed to evaluate and analyze the financial risk situation using the regression analysis of support vector machine. Thirty-six financial institutions in the 2012-2021 interval are selected as research objects for empirical analysis. Under the impact of three major events in 2013, 2015 and 2020, the stock returns of financial institutions have obvious clustering characteristics. China’s financial risk spillover index keeps fluctuating within the range of 35% to 55% from 2012 to 2021. The financial risk assessment shows that the model regression values in 2017, 2019, 2020 and 2021 are in the interval of [0.3, 0.7], and China’s finance is in a high-risk state. 2013, 2014, 2015, 2016 and 2018 are in a low-risk state.
