Abstract: Based on the cointegration theory and the fractal market theory, this paper constructed a new quantitative investment strategy for commodity futures based on information fusion and strategy switching. It verified the effectiveness and robustness of the strategy by empirical analysis. Price linkages are common among commodities. Previous studies have constructed statistical arbitrage strategies for commodity futures based on price linkages. However, this paper used price linkages to construct trend-following strategies based on information fusion. Moreover, it designed a strategy switching regime based on the fractal market theory, i.e., using trend-following strategies in trending markets and switching to statistical arbitrage strategies in mean-reverting markets. The empirical results show that information fusion can improve the performance of trend-following strategies, while applying strategy switching can further improve the investment performance. Furthermore, this paper uses Monte Carlo simulation to verify the robustness of the empirical results, and gives optimal trading strategies in different model assumptions.

摘要： 在协整理论和分形市场理论基础上，构建一种新的基于信息融合和策略转换的商品期货量化投资策略，并通过实证检验了该策略的有效性和稳健性。部分大宗商品间存在价格联动关系（同涨同跌或A涨B跌）。以往研究基于价格联动，设计并验证了商品期货的跨品种统计套利策略，而本文则利用价格联动设计并验证了基于信息融合的趋势跟踪策略。在此基础上，基于分形市场理论，增加了策略转换环节：在趋势性市场使用趋势跟踪策略，在均值回复性市场使用统计套利策略。实证结果表明，信息融合能够提升趋势跟踪策略的表现，而应用策略转换能进一步提升投资绩效。最后，使用蒙特卡洛模拟验证了实证结果的稳健性，并给出了不同模型假设下的最优交易策略。

关键词: 信息融合, 策略转换, 商品期货, 量化投资