Simulation of Enterprise Financial Asset Allocation and Stock Price Collapse Risk Model Based on DNN

With the transformation and upgrading of China's economy, the phenomenon of enterprise financialization has become increasingly prominent. Enterprises use more capital to allocate financial assets, so they rely more on the financial market rather than physical production to obtain income. Driven by the pursuit of profits by capital, a large amount of capital flows to the virtual economy, and enterprises gradually deviate from their main business, which is easy to accelerate the formation of foam in the capital market price. At the same time, enterprises can also use financial assets to cover up their real performance. The potential operational risk increases, and the risk of stock price collapse increases, which may affect the stable operation of China's stock market. A risk behavior prediction model based on deep neural network(DNN) is designed. This model identifies high-risk traders by analyzing massive trading data. The AUC value and KS value of the network model based on long-term and short-term memory depth proposed in this paper are slightly higher than those of XGBoost model, and compared with XGBoost model, it has the advantages of low parameter adjustment cost and easy iteration. On the whole, the model proposed in this paper can replace XGBoost as a new financial risk control model after a period of online parallel verification.