Commodity Futures Price Conditional Var-based Risk Analysis

VaR is a major tool to assess investment risk nowadays. The traditional core of study of VaR focuses on the distribution of the return series, much of which is to use the time series as the tool to dig and mine the maximum information of return series. But the author adds the price level to the modeling of return series basing on the hypothesis that the return series has the relation with the price level.In the study of future price risk based on the future contracts of sugar of Zhengzhou Commodity Exchange, the absolute value of price can't reflect the distance of deviation from the fluctuation core of axis because the price series are not stationary. So the author introduce the BIAS technical analyze indicator into the modeling of return series to reflect the influence of price level. After analysis of comparison, the author finds the standard deviation of residual of established model into which BIAS is brought is the least when the parameter of BIAS equals to 6.Then the author establishes all available sugar contracts with the parameter of BIAS equaling to 6.The consequence shows that the information of return series can be well explained by the introduction of BIAS into the modeling of return series and most of established models don't have heteroskedasticity effects and the model to measure VaR based on the price level is simple and efficient.In the study of future arbitrage risk based on the future contracts of copper of shanghai futures Exchange, the author respectively uses two arbitrage methods which are parity arbitrage and spread arbitrage since the price series are stationary but non-normal. During the arbitrage process using parity arbitrage method, the return series dynamic conditional normal distribution is established mainly according to the lag residuals of established regression model between price series of future contracts, and then the VaR to forecast can be calculated. After establishing all available contracts of copper using parity arbitrage method, it can be found that the return series distribution can well forecast the value of VaR and the established models don't have heteroskedasticity effects. And then, an evaluation of the effectiveness of measurement of VaR to all related arbitrage contract is done according to established model. Meanwhile, the assessment of investment value of all related arbitrage contract is done according to the risk assessment indicator. During the arbitrage process using spread arbitrage method, the subtraction of price of related arbitrage contracts is analyzed to establish model mainly taking stationary time series modeling method. It is found that the model using parity arbitrage method can forecast VaR well according to the lagged subtraction of price which can make the return series subject to the dynamic conditional normal distribution. According to the comparison between the two arbitrage methods based on the price condition, it can be known that the model established by spread arbitrage method is simple and easy on the investment position, so spread arbitrage method is proffered in the calendar spread arbitrage. In the study of future arbitrage risk based on the price series are stationary and normally-distributed, the formula of VaR calculation mentioned in reference article is used to calculate VaR. In this article, the moving average method is taken. Theμ,ρandσare treated as variables. The calculation ofμ,ρandσis from the sample of price before a certain time which counts up to n. It is found that when n increases, the calculated VaR not also overestimate the risk but also increases the failure rate obviously. When n equals to 6, the consequence is better than 12 and 24.In the study of Hedging risk based on the future contracts of cotton of Zhengzhou Commodity Exchange, the traditional OLS hedging model and univariate GARCH hedging model all can't pass the parameter significant test, and error correction model is also unfit since there is no cointegration relationship between the spot price and future price. When the first order lag of basis is introduced, the hedging model passes through the parameter significant test and the forecast of VaR according to the hedging model is all right. Therefore, the last basis has influence to the change of spot price and future price. It is demonstrated by the introduction of lagged basis during the modeling.