Estimation And Applications Of Semi-Parametric Networks Time Series Model

As a new method of analysis,network analysis techniques are applicable to many fields,such as finance,social science and even neuroscience.This paper is based on the network analysis technology to "Shanghai 50" 50 stock price fluctuation network model.Firstly,the vector autoregressive model is used to model the stock data,and the mutual influence coefficient is obtained.Then the regenerated kernel function is added to the model to fit the common influence of external factors such as the market index on all stock price fluctuations.Finally,a semi-parametric network time series model is obtained,by which the sparse connection relationship between stocks can be obtained.In order to obtain the influence coefficient between stock price fluctuations,this paper proposes an estimation method based on the semi-parametric network time series model.Numerical simulation is used to verify the correctness of the method and program.Finally,this paper also USES the semi-parametric network time series model to predict the volatility of 50 stocks in "Shanghai stock exchange 50" in a rolling manner,and finally finds that the effect is better than the prediction result of LSTM.