基于网络关注度的大宗商品市场预测研究

随着大宗商品市场化的加快和电子信息技术的快速发展,以互联网为载体的网络信息将方便快捷地传递到市场及市场参与者.本文从海量开源数据出发,利用搜索引擎平台,提取核心信息构建网络关注度指标,并提出了基于网络关注度的大宗商品价格预测模型.通过引入具有不同核函数的支持向量回归模型,分别建立了针对单个市场(原油、铜以及玉米)的网络关注度预测模型和综合考虑市场间联动性的多市场网络关注度预测模型.实证结果表明,网络关注度对于市场价格的变动有显著的格兰杰因果关系,引入网络关注度指标和相关市场信息能显著提高预测精度.

Prediction of bulk commodities based on Internet concerns

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.