Research On Portfolio Risk Measurement And Optimization Based On GAS Framework

In recent years,one of the most important innovations in the field of econometrics is the proposal of the generalized autoregressive score(GAS)theoretical framework.The theoretical framework updates the time-varying parameters by constructing the score function,which provides a new theoretical framework for the research of financial time series.Its adaptability can be applied to the research of many problems in the financial field,especially in the risk research field.The risk measurement of financial assets has always been the focus of financial time series research.Accurate risk measurement and accurate risk prediction have important theoretical and practical significance for understanding the risk situation of financial assets and making relevant investment decisions.In recent years,with the rapid development of China's financial market,the number of A-share listed companies continues to increase,and investors' choice of targets also continues to increase.At the same time,the development of the futures market has also brought more financial tools for investors,and the portfolio has also extended from a single stock portfolio to the spot and Futures Portfolio.Therefore,How to choose the risk measurement method of portfolio has also become a key issue in the field of academic practice in China.In addition,with the development of computer technology,the improvement of computing power not only provides strong support for the modeling of high-frequency data,especially in recent years,the rapid rise of quantitative investment has driven the development of high-frequency trading.High frequency data can bring more information,but due to the influence of market microstructure,there is also a lot of noises.How to take appropriate noise reduction methods to build an effective high-frequency multi asset volatility model to measure the risk of portfolio has been gradually paid attention to with the development of quantitative investment technology.Therefore,based on a detailed review of the existing research results in GAS theoretical framework,risk measurement of investment portfolio,investment portfolio optimization,volatility research of high-frequency asset portfolio and optimization of futures and spot investment portfolio,this paper constructs a risk measurement and optimization model for single asset and multi-element asset portfolios based on the GAS framework.In terms of data frequency,it includes the research on the prediction of realized variances of investment portfolio under high-frequency data.In terms of portfolio asset types,the research on the optimization of hedging of futures and spot investment portfolio is also included.Specifically,this paper first introduces the GAS framework into the risk research of financial assets,gives the basic expression of the GAS framework model,improves on the basic expression,and combines it with the traditional GARCH model to build the GAS-GARCH-sst model,and analyzes the advantages of the GAS framework in risk measurement and VaR prediction through empirical research;Subsequently,this paper will single financial The measurement of asset risk extends to the research of multi asset portfolio.In order to depict the dependence of assets in the portfolio,the R-Vine structure is constructed,and the GAS framework is introduced to propose the R-Vine-GAS-sst model,which is applied to the research of portfolio based on China's stock market;Step by step,the research is expanded to the research of high-frequency data,and the GAS framework is applied to construct the GAS-Wishart-GARCH model,which provides a new high-frequency multi asset volatility model;Finally,the paper further expands the investment portfolio from a single multi head portfolio to a spot investment portfolio with multiple space and two sides.Based on the principle of risk minimization,the GAS framework is combined with the extreme value theory and the Copula function.In order to optimize the hedging effect,a variety of hedging estimation models are established.The main conclusions of this paper are as follows: First,introduce the GAS theoretical framework into the risk research of financial assets,and propose an improved model that can effectively solve the problem of outliers.Therefore,it can be seen that the introduction of a score function as an update-driven GAS framework The innovative GAS-GARCH-sst model can reduce the influence of outliers on the model and make the risk measurement of financial assets more accurate;Second,compared to the traditional GAS-GARCH-sst model,the innovative and complete GAS-GARCHsst model based on the GAS framework has stronger ability to predict the VaR of a single asset,and the application of the GAS framework can effectively improve the risk prediction ability;Third,in terms of portfolio risk measurement,the GAS theoretical framework can significantly improve the predictive ability when applied to the risk prediction of the Chinese stock market portfolio,and the R-Vine-GAS-sst model can significantly improve the prediction effect of portfolio VaR;Fourth,in terms of portfolio optimization,the new portfolio optimization model based on the GAS framework can achieve a higher level of return while reducing risks with a higher degree of optimization;Fifth,in the minimum risk portfolio model,using CVaR as a risk measurement method has a better effect than VaR;Sixth,in the study of financial high-frequency data,the volatility model based on the 5-minute sub-sampling noise reduction technology has a better effect.In the Chinese market,it is estimated that a better noise reduction method of the covariance matrix has been achieved;Seventh,the GAS-Realized-GARCH model based on the GAS framework can achieve a better prediction effect on the volatility of high-frequency multi-assets,and at the same time through and practice a comparison of the EWMA model commonly used by investors and regulators in the field found that the GAS-Realized-GARCH model can quickly respond to changes in the realized covariance,which is more able to adapt to change in risk and predict more accurately;Eighth,The application of GAS framework has obvious advantages in investment portfolio hedging optimization.In the out-of-sample interval,the optimal hedging ratio estimated by the extreme Normal-Copula-GAS and extreme Frank-Copula-GAS models can effectively reduce the risk of the CSI 300 ETF and futures portfolio;Ninth,the naive hedging strategy with the same proportion of futures and spot stocks achieves a significantly lower risk reduction than the OLS,DCC-GARCH,and extreme Normal-Copula-GJR,extreme Frank-Copula-GJR,extreme Normal-Copula-GAS and extreme Frank-Copula-GAS models,and the cost is higher,and should not be used in investment practice.