For the possible economic crisis of real estate companies, by analyzing the shortcomings of support vector machine (SVM) model, after optimization with particle swarm optimization (PSO), the PSO-SVM model was established by changing economic condition parameters and using data to warn the real estate economic crisis. Then, this model is used to warn the economic capacity of four real estate companies in Beijing. The results show that this model can further predict and analyze the solvency index, operating ability index, development ability index, profitability index, cash flow index, and comprehensive financial index of real estate companies and come to the conclusion that profitability index is the most important, and higher profitability can resist the harm brought by economic crisis. Finally, it analyzes the economic crisis that real estate companies may encounter and puts forward specific suggestions.

At present, the economic development of the world’s major economies is showing a positive and positive state. Driven by the development of related industries, the development of the financial field is also changing with each passing day. Various activities in the financial industry are in full swing, and the forecasts of related prospects are also full of uncertainties. Summarizing the laws of financial activities through technical means and making accurate predictions of future trends and trends is a hot research direction that relevant researchers pay attention to. Accurate financial forecasts can provide reference for financial activities and decision-making to a certain extent, promote the steady development of the market, and improve the conversion rate of financial profits. As an algorithm model that can simulate the biological visual system, the convolutional neural network can predict the numerical trend of the next period of time based on known data. Therefore, this paper integrates the support vector machine with the established model by establishing a convolutional neural network model and applies the prediction model to the prediction of financial time series data. The experimental results show that the model proposed in this paper can more accurately predict the trend of the stock index.