An improved on BP neural network for financial crisis prediction

Different from the previous study, this article, taking full account of China's current situation of accounting, information supply, constructs a six-category warning index system. The system constitutes index reflecting the solvency, assets and liabilities management, profitability, growth,cash flow and the condition of existing accounting information offering. In addition, given that the Shanghai and Shenzhen stock exchanges use listed company's financial situation of the year (t-1) to determine whether to give the company special treatment in the year, it is useless to forecast using data from one year before the crisis happen. This article tests the effect of the forecasting model of the improved BP neural network, sampling ST Companies' data of two years, three years and four years before they were special treated. The results show that BP neural network model has an accuracy rate of 88.5% two years before the financial crisis of, with obvious advantages and value.