Risk Measurement Research Of Steel Futures Based On Extreme Value Theory

Due to the globalization of economy and the rapid development of financial markets,the volatility of market has become more and more aggravated,financial events occur more and more frequently,therefore the financial risk management has become more and more important.Extreme events once occur,the loss it brings will be very huge.It can cause the bankrupt of the financial institutions.Even worse,one country will break down and global economy crisis will break out.In March 2009,rebar futures launched in the hope of hedging.Only the risk is accurate measured and effectively managed,can ensure the financial markets smoothly and orderly develop.Currently,VaR and CVaR risk measure method are generally used.In the theoretical part,the paper introduces several traditional risk measurement methods.Most of these methods assumed financial series obey a certain distribution,but a large number of empirical studies have shown that the actual market does not obey these distributions.This assumption has a risk of model assumption,may cause the actual risk to be underestimated.Extreme value theory overcomes these shortcomings.It does not assume that financial data obeys a certain distribution and focuses on the characteristics of the tail distribution.Generally financial data is volatility clustering,so we use GARCH models to describe this phenomenon.Finally,we combine GARCH models with extreme value theory models to construct a new model GARCH-GPD model.It uses GARCH models to filter financial data and use Pareto distribution to fit the residuals,then we can compute the risk value.This model takes into account both volatility clustering and the extreme risk,so it's more acurrate and effective than other traditional models.In the empirical part,we select steel futures closing price of March 27,2009 to December 31,2015 for research.After descriptiving and checking the steel futures rate of return,found that the rate of return does not obbey normal distribution and has a fat tail.Therefore,we use GARCH models to measure the risk.As we are more concerned about the occurrence of extreme events,consider using the extreme value theory models measure China's steel futures market risk.Finally,we use GARCH-GPD models to compute the risk and compare with other models by backtesting and Kupiec test.Although GARCH-GPD model is more conservative,but more acurrate.Thus we can control the risk of futures market more effectively.