二 〇 二 五 年 度 2025 年 11 月 28 日 从信念到模型验证：估值与周期双轮驱动 ——大宗商品中观轮动系列（二） | 虞堪 | 投资咨询从业资格号：Z0002804 | yukan@gtht.com | | --- | --- | --- | | 邵婉嫕 | 投资咨询从业资格号：Z0015722 | shaowanyi@gtht.com | | 李翔云（联系人） | 期货从业资格号：F03149627 | lixiangyun@gtht.com | 报告导读： 商品中观轮动的研究初衷，是将"主观+量化"理念有机结合并付诸实践。我们希望尽可能降低策略 中因子、参数及模型的特异性，注重策略的可解释性及可归因性，给出经得起时间验证的模型。我们构建 的月频品种簇中观轮动模型，样本内平均年化收益率 17.79%，夏普比率 1.44，回撤-5.70%，月度胜率 68.98%；样本外 2025 年 1-11 月总收益录得 15.43%，回撤录得 1.09%，月度胜率 70%，仅三个月录得负 收益率。 本篇报告是中观轮动系列的模型搭建部分： 请务必阅读正文之后的免责条款部分 1 国 泰 君 安 期 货 研 究 ...

Report Overview - The report focuses on the meso - level rotation of commodities, aiming to combine "subjective + quantitative" concepts. It provides a theoretical foundation for subsequent model building [1][63]. Industry Investment Rating - No industry investment rating is provided in the report. Core Views - The report emphasizes the construction of a dynamic cognitive system for investment. It analyzes the rotation phenomena and mechanisms in the equity and commodity futures markets, and constructs a research framework for macro - and fundamental - valuation rotation in the inventory cycle. It also quantifies the meso - level rotation targets as commodity "variety clusters" [1][63][64]. Summary by Directory 1. Significance of Meso - level Research - In the financial market, a dynamic cognitive system is needed for investment. Since 2022, the Chinese commodity futures market has changed, with lower volatility and reduced effectiveness of factors. Research on meso - level "commodity collections" can avoid co - decline risks, capture structural opportunities, and identify potential trends [3]. 2. Meso - level Rotation in the Equity Market 2.1 Rotation Phenomenon in the Equity Market - The size premium and value premium in the Fama - French three - factor model are core factors for explaining stock return differences, providing a theoretical basis for style rotation [4]. 2.2 Formation Mechanism: Cycle Alternation and Capital Game - **Cycle Alternation (Top - down)**: Style rotation in the equity market stems from the cycle of the economic cycle. Different stages of the economic cycle lead to different dominant styles, such as small - cap and growth styles in the early recovery stage, and large - cap and value styles in other stages [10][12]. - **Capital Game (Bottom - up)**: Style rotation is driven by the game between existing and marginal funds. Existing funds lead to style differentiation, marginal funds strengthen the style, and style conversion occurs when the valuation deviates from the fundamentals [15][16]. 3. Meso - level Rotation in the Commodity Futures Market 3.1 Rotation Phenomenon in the Commodity Futures Market - By analyzing the rotation speed, intensity, long - short suitability of the first and last positions, and the distribution of the first and last positions of commodity futures market indices, it is verified that there is a rotation phenomenon in the commodity futures market. The first - place average return is 5.79%, the last - place is - 4.43%, and the average difference is 10.22% [22][26][29]. 3.2 Formation Mechanism: Game between Reality and Expectation in the Inventory Cycle - The meso - level rotation in the commodity futures market is driven by the transfer of the main contradiction in the inventory cycle. Different stages of the inventory cycle have different logics, such as "reality - driven, expectation - following" in the passive de - stocking stage and "expectation - driven, reality - pressured" in the passive re - stocking stage [30][33][34]. 3.3 Dynamic Framework: Rotation of Macro - financial and Fundamental Valuations - A preliminary research framework for macro - and fundamental - valuation rotation in the inventory cycle is constructed based on Bayesian thinking. The reality side is represented by fundamental valuation, and the expectation side is represented by macro - valuation [40][44]. 4. From Sector to Variety Cluster Rotation - Sector indices have limitations, so variety clusters are introduced. By considering the industrial chain and return clustering, 16 variety clusters are divided, including those in the black, non - ferrous, energy - chemical, agricultural, and precious metal sectors. The variety clusters have lower correlation and better risk - dispersion properties [49][57][60]. 5. Summary - The report combines "subjective + quantitative" concepts. It analyzes the rotation phenomena and mechanisms in the equity and commodity markets, constructs a research framework, and divides variety clusters, providing a theoretical basis for subsequent model building [63][64][65].