Search

With the rapid development of bulk commodities and electronic technology, the network information carried by the Internet delivers quickly to the market and the participants in it. Using search engines that equip with massive open-source data, we propose in this paper a prediction model of the price of bulk commodities, by constructing Internet concern indices from the key searching information. Due to the support vector regression (SVR) model with different kernel functions, we build a prediction model respectively for the single market of crude oil, copper and corn. In addition, considering the co-movement among commodity markets, we further present a model with Internet concerns in terms of multiple markets. Empirical results demonstrate that the Internet concerns present a significant Granger causality on the variation of market price. Meanwhile, taking into account the Internet concern indices as well as information from related markets can improve the prediction accuracy in a remarkable amount. [ABSTRACT FROM AUTHOR]