Clearing houses or huge broker-dealers usually apply portfolio approach for the estimation of traders’ liabilities and restriction of their risk. It supposes that they analyse possible changes of portfolio’s value within a given time horizon. For clearing houses (in USA as well as in Europe) there exists a standard methodology called SPAN, developed by Chicago Mercantile Exchange, where scenario approach is implemented. Sometimes this approach fails due to the finite number (16) of considered scenarios. It seems that parametric approach provides better quality, but in this case man needs a fast algorithm for the Implied Volatility (IV) curve forecast. By now there exists an algorithm which uses a polynomial approximation for the IV curve merged with Kalman Filtering procedure as a forecasting tool. We develop a new model and provide statistical evidence which shows that approximation of IV curve by a special class of functions give better results. Correspondent analysis is based on the data from US options market.