There is no doubt that probability distribution is primary and important for the risk analyses on financial time series. And various non-Gaussian distributions have become one of focused and unsolved problems, especially for those studies on the real continuous variables. So this paper concentrates on the intelligent algorithm for probability density estimation by Least Squares Support Vector Machines (LS-SVM), and its application on the electricity price. Moreover a practical probability density modeling of electricity price is implemented by LS-SVM. Finally, case studies on the electricity price of New England electricity market have proved the validity of the proposed model.