This paper examines the crush arbitrage of soybean and rapeseed futures, which is based on inter-commodity arbitrage in the oil processing process. The paper takes soybean, soybean meal and soybean oil futures listed on the Dalian Commodity Exchange, rapeseed, rapeseed meal and rapeseed oil futures listed on the Zhengzhou Commodity Exchange as research objects. Based on the statistical arbitrage model, empirical studies on two kinds of crush arbitrage are conducted. The improvement of this paper is to examine the characteristics of price volatility from the perspective of time-varying fluctuations, and then improve the setting of trading thresholds. The empirical results show that the use of GARCH model to solve the time-varying variance and set the feasibility of the threshold improvement, especially, rapeseed crushing arbitrage strategy portfolio returns better.