中文摘要：

国内压榨市场大豆主要来源于国外进口,并且国内压榨市场缺乏对大豆价格制定的主导权,故可通过大豆及其压榨品豆粕、豆油进行跨商品套利.综合运用协整检验、误差修正模型（ECM）研究大豆、豆粕、豆油三者价格之间长期存在的相互关系,同时通过设置不同开平仓阀值,对比均值回归模型套利策略和Elman神经网络模型套利策略,实证结果表明：我国大豆期货及豆粕和豆油期货三者进行跨商品套利可行,能够获得正向的套利收益率;在不同的开平仓阀值下,Elman神经网络模型较均值回归模型能够得到更好的套利结果.

英文摘要：

This paper studies on the intra - commodity arbitrage of soybean futures and its squeezed soybean meal, soybean oil. In this paper, we use co -integration test and Error Correction Model to researd the relationship among the futures prices of soybean, soybean meal and soybean oil. By setting thresholds of opening and closing positions, this paper compares the arbitrage results of equilibrium regression method and those of Elman neural network method. The empirical results show that under the different thresholds of opening and closing positions, the Elman neural network method leads to better arbitrage results than equilibrium recession method.