Time Series Clustering In Portfolio Risk Management

After the financial crisis of2007,the international economic and financial markethave changed a lot.the increased volatility and continuing turmoil in financial marketsare more prominent aspect.Since China joined the WTO,constantly faced with thepressure of capital market liberalization.International capital markets growly affectthe domestic capital market,leading to increased risk of domestic financial markets.Be-sides,increased competition has also led to financial market risks increase in recentyears.Thus,people pay more attention to managing financial risks,and are increasinglyconcerned about the risk of investment.At investment theory and practice,Markowiz mean-variance model provedes abasic and complete framework for the investment decisions.And it become one of themainstream methods of financial analysis.In the mean-variance model,the core part isthe measure of risk,However,the model presences a significant defect by usingvariance to measure risk,which require to explore a more accurate risk measurementmethods.Besides,pricing theory and modern investment theory are built on the basisof risk metrics.From this,we can see the risk measure in the financial sector occupiesan important positon.Thus, in the past60years,from the variance to VaR,people havebeen finding a more rational risk measurement methods.Firstly,we analyze VaR andCvaR from the evolution of the current prevalence,characteristics and consistencyaxiom,and compare their respective strengths.Finally,we choice CvaR to build theportfolio.When building the portfolio,the paper use the time series clustering methodby clustering CvaR sequence to make a minimum overall portfolio risk,and thenchoose the least risky portfolio.It will guide significancely for practical investment.