Abstract: Coal resource plays a significant role in primary energy production and consumption in C hina. Thus, coal price has a great influence on national economy, it is very meaningful to predict coal price. However, because of limited data availability, further study is requied to investigate the precision and reasonability of those methods such as multiple regression method, dynamic analysis model, neural network method and so on. In this study, we use Artificial Neural Network (ANN) to select key influencing factors of coal price. And then, we introduce the Bacterial Colony Chemotaxis (BCC) algorithm based on random Nelder Mead (RNM) to determine the extra-parameters used in least squares-support vector machine (LS-SVM) for coal price prediction rapidly and reasonably. At last, a case study of D atong premium blend coal at Qinhuangdao port is presented predicting its coal price during the twelfth Five-year-planning. Compared with the prediction results of ANN and BCC, the suitability and novelty of ANN and RNM-BCC-LS-SVM is fully demonstrated.