It is one of the most important tasks of risk controlling in the financial market. The margin trading plays a leverage way in the future's trading while the trading risk is amplified in the same way. So it is so important of the Margin setting that the level of corporately Margin will affect the liquidity of the market even damage the efficiency of the market.Therefore the determination of reasonable margin level is necessary. On the question of that, in order to get the conclusions of expectations, the treatise gives the theoretical research with modern risk controlling model and does the empirical study on the bean oil and dregs of beans all of which have the maximum degree and being traded on the Dalian Commodity Exchange.In order to compare the margin with group variety, the treatise analysis the optimal margin level of Single variety firstly with the measure of VaR. In the fitting of single variety yield distribution, the time-varied GARCH model is drown in the conditions of heteroscedasticity. From the result of the empirical study, for the measure of single variety, it is easy to make the margin all lose in the model of Residual GED distribution in the computation of VaR.Drowning the Copula function to fit the marginal distribution, the treatise uses the Unsymmetrical Laplace distribution and Garchs distribution to fit the data, all of which is the balance between the model's reliability and the simulation results. In the measure of group variety margin, the Monte Carlo stochastic modeling method is taken into consideration. Generally speaking, the difference between the two methods is not significant. It is not the simply plus-minus for two-related varieties with the inter structure. From the study, we can see that there is less risk for the group variety on the base of Copula function, and the level of 1% margin nearly covers all the risk under 99%confidence level.After the finish of normal risk measurement, the treatise turns the aim to Extreme Value Theory with the model of POT in the distribution of General extreme distributions (GEV), and the generalized praetor distribution (GPD). From the result we know that the higher margin level still needed to eliminate the Extreme Value under the higher confidence level. The higher reserve and the normal margin nearly amount to 10% which accounts for why the margin of future companies in our country nearly all above 10%. The 10% margin can nearly eliminate the Extreme Value after the covering of normal risk. The only difference is the margin contains the normal margin and the reserve. |