Abstract: The research on demand forefoundry for supply chain at home and abroad can be roughly divided into linear and nonlinear method. As nonlinear method can accurately forefoundry the random fluctuation of practical demand, it has become a hotspot in research on the demand forefoundry for supply chain. The Local Area Single-Order Weighting Forefoundry, Maximum Lyapunov Index Forefoundry and Macrocosm Support Vector Machine Forefoundry are used most frequently. Through the empirical comparative study, it proves that, the method of Support Vector Machine (SVM)based on the space reconstructive theory can forefoundry the fluctuating trend of practical demand with high degree of accuracy.