An Empirical Study On Extreme Risk Early Warning Model Of Stock Market Based On Artificial Intelligence

With the development of economic globalization,extreme risks occur frequently in the stock markets of various countries,causing great harm.Therefore,the establishment of a reasonable stock market extreme risk warning mechanism is particularly important,with a high practical significance.However,the stock market is a complex whole,which is affected by many aspects."abnormal" phenomenon often occurs in the market,and the data show a complex non-linear relationship,which will adversely affect the establishment of extreme risk warning model.In this context,the artificial intelligence method has been favored by more scholars,and it can deal with the non-linear problems between data.Deep learning is one of the fields concerned by scholars in the field of artificial intelligence.At present,there are few domestic researches on stock market extreme risk warning based on deep learning technology.In this paper,the deep learning full connected neural network(DNN)and long and short term memory network(LSTM)are introduced into the stock market extreme risk warning,and a stable and effective model is established for the early warning of extreme risks in China's stock market.In this paper,the csi 300 market is taken as the research object.Firstly,the multi-fractal method is used to define the characteristic indicator variables of the state of the stock market.The risk characteristic indicators are screened from both inside and outside the market.Then,an extreme risk warning model of DNN and LSTM stock market was established for analysis and evaluation,and the model was applied to the extreme risk warning of the csi 300 market.The research shows that DNN and LSTM extreme risk warning models show good stability in China's stock market,and both of them can carry out scientific and effective early warning for extreme risk in China's stock market,which is of high practical significance.