Research On Stabilization Measures And Prediction Method Of Stock Market Based On Complex Network

It is believed that the stock market could be regarded as a complex nonlinear dynamic system with numerous variables.The applications of complex networks have provided a new perspective for studying the in-depth mechanisms of the stock market,such as the operation mechanism,evolution characteristics,policy implementation and trend forecast.It has positive significance for improving market supervision and promoting the steady and sound development of the stock market.In this study,stock market networks are proposed from Chinese stock market.Lately,the effectiveness of stabilization measures and a prediction method of the stock market networks are deeply studied based on the stock market networks.Main contents and innovations are as follows:(1)The stock market complex network models are constructed based on the Pearson correlation analysis and the fractal analysis.Statistical characteristics and topology structures of the stock market complex network models are studied based on the Pearson correlation analysis.To overcome the limitations of Pearson correlations on nonlinear and non-stationary time series,a time-migrated DCCA cross-correlation coefficient is proposed based on the fractal analysis.The time-migrated DCCA cross-correlation coefficient is suitable for non-stationary time series and detecting the time-migrated correlations,which could ensure more relevant results than the DCCA method.The stock market complex network models are constructed based on the time-migrated DCCA cross-correlation coefficient.These models are more suitable for further investigation on internal properties and external influencing factors of the stock market.(2)The effectiveness of stabilization measures on stock market network models is studied.The dynamic evolution,stability and systemic risk of the stock market are studied based on complex network simulation experiments.According to the experimental results,corresponding policy suggestions are put forward from aspects of necessity,effectiveness,moderateness and pertinence.The purpose of this model is to provide a theoretical basis for market supervision on risk prevention and stabilization measures implementation of the market crash.(3)The hybrid method is proposed to predict the variations of five stock prices in the securities plate sub-network.This method integrates independent component analysis and multivariate long short-term memory neural network to analyze the trading noise and improve the prediction accuracy of stock prices in the sub-network.The algorithmic framework of this hybrid method is decomposition,noise reduction,training and reconstruction.This method reduces the calculation amount and improves the prediction accuracy for trend predictions of the stock market network models.The experiment results indicate that the hybrid method outperforms the benchmark approaches,especially in terms of the stock market complex network.This paper takes complex network technology as the theoretical basis and focuses on the stock market network model construction,stabilization measures and prediction method of the stock market networks.The research results can help supervisors of the stock market to evaluate the market situation and improve the efficiency of policy implementation,which has positive significance for increasing market stability,reducing market risk and achieving steady and sound economic development.