Quantification And Analysis Of Financial Market Uncertainty Based On The Deep Learning (Seq2Seq) Model

From the point of view of the concept of uncertainty,there are many viewpoints on the definition of the concept of uncertainty.The understanding of different uncertainties also leads to differences in the quantitative and quantitative methods of uncertainty.For China,the stable operation of the financial market is related to whether economic growth can continue to operate steadily.Under the background of globalization,China's financial market is gradually opening the pace of opening up to the outside world.International and domestic economic and financial uncertainties will affect China's financial market.By constructing China's financial market uncertainty index,it can play a warning role in the operation of financial markets,providing a more intuitive reference for financial policy makers,so that the macro-control of the authorities can be relied upon and thus Improving the efficiency of economic and financial decision-making,strengthening market supervision and supervision,and stabilizing the development of the financial economy have great practical significance.At present,domestic literature mostly focuses on the micro-level financial financial indicators(VAR,SES,MES,probability of default risk,etc.)to measure the marginal contribution of microfinance institutions to the overall system financial risk,and then to identify and divide the system.Financial risks provide an effective basis.However,the micro-level indicators can not answer the macro-wide problem,it is difficult to identify and monitor the dynamic changes of the overall system financial risks,and thus can not provide an effective reference for China's macro-financial prudential supervision,policy formulation and adjustment.Modern economic operation generates a large amount of economic variable data.Uncertainty is an important factor in economics.The key to measuring economic financial uncertainty is to clarify the relationship between uncertainty and economicand financial variables,and how to Economic uncertainty is measured in economic and financial data.Therefore,this paper will construct a deep learning model from the perspective of large-dimensional data,measure the financial uncertainty index,and then study the uncertainty of China's financial market and the source of uncertainty.In the model construction of the financial uncertainty index: First,by quantifying the uncertainty index of China's financial market in high-dimensional economic and financial variables,it has many advantages over the use of other substitution variables instead of the uncertainty index;This method is closer to the definition of uncertainty,which can capture the uncertainty of different aspects of the financial market more widely,and then use the higher-dimensional data layer to analyze the source of uncertainty.Second,by transforming the sequence-to-sequence model(Seq2Seq)model in deep learning,the prediction accuracy of the prediction model in the financial uncertainty index can be significantly improved,and the prediction residual of the model can be reduced.This kind of improvement of the accuracy of economic and financial variables,compared to linear regression models or autoregressive models,the deep learning model has a stronger ability to eliminate the predictable components of economic and financial variables and ultimately quantify the uncertainty..However,deep learning has its unstable factors,and the prediction accuracy of the model will not increase with the increase of the complexity of the model.The empirical study found that deep learning can be used in shallow models(0to 4 layers).To achieve the optimal prediction effect,and increase the complexity of the model,not only the prediction accuracy will not be reduced,but the model will not converge.Thirdly,in the different periodic predictions of the constructed financial uncertainty index,no obvious lags were found in different periods;in different prediction periods,the forecasting trends in the same time period are basically the same.Therefore,the predicted value of the financial uncertainty index constructed by this model can provide a reference for future China financial market uncertainty index or expectation.In the empirical analysis of the financial uncertainty index,this paper comparesthe relationship between the constructed Chinese financial uncertainty index and the economic policy uncertainty index in detail;discovers the uncertainty of China's financial market and the economic policies of major international economies.The uncertainty is significantly correlated.In addition,the correlation between the uncertainty index of China's financial market and the volatility of the stock market is also compared in detail.It is found that the financial uncertainty index is significantly correlated with the volatility of the main financial composite index of the Shanghai market and the Shenzhen market,and the international financial market.The volatility of the main financial composite index is not significantly correlated.It is believed that the uncertainty of China's financial market is mainly affected by the domestic economic environment.In the deep learning interpretation of the financial uncertainty index,firstly,by decomposing the data sources of the financial uncertainty index,the sources of financial uncertainty are divided into the uncertainty from domestic finance,the uncertainty from international finance,and China's macroeconomic uncertainty and industry uncertainty are four major parts.By observing the four parts of the financial uncertainty index,we can see the source of uncertainty more intuitively.Finally,the decision tree model is constructed to explain the cause of the prediction results of the deep learning prediction model,and the source of the specific index variables of uncertainty is analyzed by the importance of the variable factor of the decision tree model;among them,public utilities,communication industry,national defense industry The changes in market value of real estate and other sectors have contributed the most to the impact of financial uncertainty.