Abstract: In this paper,based on the idea of EWMA model and GARCH model,combined the idea of SPAN system,using the important functions of the margin in the future markets as the consideration factors which remedy the largest loss in tomorrow and guarantee the trade running normally,making use of venture administering methods of symbolic statistics and VaR,we establish the moving adjust model of margin.Consequently,under the precondition of determined venture coefficient, decrease the rate of margin to the best of our abilities and offer new computing method for the determination of the future markets.The characteristics of this model are as follows:Firstly,we make use of the GARCH model of VaR method to determine the key parameter and attenuation factor of EWMA model, this will solve the problem to assign the decay factors by person causing the model with strong factors man-made. Secondly, introduce the fluctuation functions into the model, use them to confirm the fluctuation coefficient following the time "t", and adopt a lot of data of soybeans and soy meals contracts to fit the fluctuation functions,thus get the continuous function between the fluctuation functions and the time"t".This model can be used to replace the coarse functions used in Dalian Commodity Exchange.Thirdly,adopt nearly 4500 data of the history contracts, and give statistical analysis to the futures contracts of soybeans and soy meals to get the probability distribution of the limit up/down situation of the two types of contracts.Finally,comprehend each situation to get the final margins,thus make the model more comprehensive.

Key words: margin model, moving model, futures trade, generalized autoregressive conditional heteroskedasticity model(GARCH), fluctuation function