申万金工 ETF 组合 202510 相关研究 证券分析师 沈思逸 A0230521070001 shensy@swsresearch.com 邓虎 A0230520070003 denghu@swsresearch.com 研究支持 白皓天 A0230525070001 baiht@swsresearch.com 联系人 沈恩逸 (8621)23297818× shensy@swsresearch.com 申万宏源研究微信服务 2025 年 10 月 10 日 请务必仔细阅读正文之后的各项信息披露与声明 宏观行业组合：针对所有标记为"行业主题"的 ETF，选择成立时间 1 年以上、当期规模 ○ 2 亿以上的产品跟踪的行业主题指数，每个月根据历史数据计算经济、流动性、信用的敏 感性得分，然后根据最新的经济、流动性、信用判断指标调整得分方向后进行加总，最终 得到排名前 6 的行业主题指数,然后取对应规模最大的 ETF 进行等权配置。当前经济前瞻 指标开始回升、流动性指标略偏紧、信用指标仍向好，组合整体转向均衡，消费比例提升。 宏观+动量行业组合：基于宏观类配置策略主要存在偏左侧的特点，胜率偏低而赔率较高, ● ...

报告导读： 风格轮动模型方面， Q3 模型均预测准确； Q4 模型发出小盘、成长信号。行 业轮动模型方面， 9 月，复合因子策略持续获得了正超额。 10 月，单因子策略、复合因 子策略推荐配置的多头行业均涵盖计算机、通信、电子行业。 风格轮动模型方面， 2025Q3 ，中证 1000 相对沪深 300 的超额为 1.27% ，国证成长相对国证价值的超额为 30.27% ，小盘、成长风格占优，两模型均 预测正确。 Q4 ，风格轮动模型持续发出小盘、成长信号。 行业轮动模型方面 ， 复合因子策略模型在 9 月仍获得了正超额，月收益率为 3.33% ，相对基 准的超额为 2.43% 。 9 月，单因子策略、复合因子策略推荐配置的多头行业均涵盖计算机、通信、电子行业。 大小盘风格轮动 Q4 配置信号。 根据 2025 年 09 月 30 日的最新数据， 2025Q4 ，双驱轮动策略得出的综合分数为 -1 ，预测信号为小盘。 本订阅号所载内容仅面向国泰海通证券研究服务签约客户。因本资料暂时无法设置访问限制，根据《证 券期货投资者适当性管理办法》的要求，若您并非国泰海通证券研究服务签约客户，为保证服务质量、 控制投资风险 ...

Group 1 - The core viewpoint of the article indicates that the market is favoring small-cap and growth styles, with the style rotation model for Q3 2025 confirming this trend [1][2] - In August, the small-cap stocks outperformed large-cap stocks with a monthly excess return of 1.34%, while growth stocks outperformed value stocks with a monthly excess return of 12.76% [1][3] - The industry rotation model showed that in August, two industry combinations achieved absolute returns exceeding 12%, with excess returns above 4% [1][3] Group 2 - The dual-driven rotation strategy for Q3 2025 indicated a signal for small-cap stocks based on the latest data as of June 30, 2025, with a composite score of -3 [2] - The dual-driven rotation strategy for Q3 2025 also indicated a signal for growth stocks, with a composite score of -5 [3] - In August, the composite factor strategy achieved an excess return of 4.38%, while the single-factor multi-strategy achieved an excess return of 4.59% [3]

- The style rotation model is constructed based on micro-level stock factors, including valuation, market capitalization, volatility, and momentum. It utilizes 80 base factors to generate 640 micro features, replacing absolute proportion division of style factors with common indices as style stock pools. Random forest is employed for rolling training to avoid overfitting, enabling feature selection and style recommendation. The framework integrates style timing, scoring, and investment implementation[3][8][9] - The performance of the style rotation model during the backtesting period (2017/01/01-2025/08/31) shows an annualized return of 17.08%, annualized volatility of 20.07%, IR of 0.85, monthly win rate of 55.77%, and maximum drawdown of -29.89%. When hedging against the market benchmark, the annualized return is 10.42%, annualized volatility is 13.03%, IR is 0.80, monthly win rate is 56.73%, and maximum drawdown is -9.57%[9][10] - The style rotation model's September 2025 timing direction focuses on growth, large-cap, momentum, and high-volatility factors[17] - The latest holdings of the style rotation model for September 2025 include ETFs such as Semiconductor Leaders ETF (159665.SZ), Big Data ETF (159739.SZ), Artificial Intelligence ETF (159819.SZ), Fintech ETF (159851.SZ), and 5G ETF (159994.SZ)[2][20]

Group 1: Report Industry Investment Rating - No industry investment rating is provided in the report. Group 2: Core Viewpoints of the Report - The report constructs multiple ETF portfolios using macro - based methods and a trinity style rotation model, aiming to capture investment opportunities in different market conditions [2][5]. - Different portfolios have different characteristics. For example, the macro - industry portfolio is adjusted monthly based on economic, liquidity, and credit conditions, and currently leans towards growth with more pharmaceutical holdings [8][11]. - The macro + momentum industry portfolio combines macro and momentum methods, and has performed well this year, almost outperforming the index every month [14][19]. - The core - satellite portfolio uses the CSI 300 as the base - position and combines different sub - portfolios to achieve relatively stable performance [20][24]. - The trinity style rotation ETF portfolio uses macro - liquidity as the core to construct a style rotation model and provides 8 style preference results [6][25]. Group 3: Summary According to Relevant Catalogs 1. ETF Portfolio Construction Methods 1.1 Based on Macro Method of ETF Portfolio Construction - Calculate macro - sensitivity of indexes tracked by broad - based, industry - themed, and Smart Beta ETFs according to economic, liquidity, and credit variables. Consider adding momentum indicators for complementarity [5]. - Traditional cyclical industries are sensitive to the economy, TMT is sensitive to liquidity, and consumption is sensitive to credit. State - owned enterprises and ESG - related themes have low sensitivity to liquidity and credit [5]. - Construct three ETF portfolios: macro - industry portfolio, macro + momentum industry portfolio, and core - satellite industry portfolio, and rebalance monthly [5]. 2.2 Trinity Style Rotation ETF Portfolio Construction - Build a medium - to long - term style rotation model centered on macro - liquidity, and compare it with the CSI 300 index. - Construct three types of models: growth/value rotation model, market - capitalization model, and quality model. Combine the results of the three models to get the final style preference, with a total of 8 style preference results [6]. 2. Macro Industry Portfolio - Select industry - themed indexes tracked by ETFs with a listing period of over 1 year and a current scale of over 200 million. Calculate sensitivity scores of economy, liquidity, and credit monthly, adjust the score directions according to the latest indicators, and sum them up. Select the top 6 industry - themed indexes and allocate equally among the corresponding largest - scale ETFs [8]. - Currently, due to the economic downturn, slightly tight liquidity, and good credit, it selects ETFs insensitive to the economy and sensitive to credit, leaning towards growth with more pharmaceutical holdings [11]. 3. Macro + Momentum Industry Portfolio - Combine macro and momentum methods. Use clustering to divide industry - themed indexes into 6 groups, and select the product with the highest 6 - month increase in each group for equal - weight allocation [14]. - Both the macro and momentum parts select many pharmaceutical - sector products, and the gaming and Internet sectors also account for a large proportion. The portfolio has performed well this year, outperforming the index almost every month [16][19]. 4. Core - Satellite Portfolio - Design a "core - satellite" portfolio with the CSI 300 as the base - position to address the high volatility and rapid industry rotation of industry - themed ETFs [20]. - Build three sub - portfolios: a broad - based portfolio, an industry portfolio (using the macro + momentum industry portfolio), and a Smart Beta portfolio. Weight the three sub - portfolios at 50%, 30%, and 20% respectively to get the final portfolio. The portfolio has performed stably this year, also outperforming the index almost every month [21][24]. 5. Trinity Style Rotation ETF Portfolio - The model currently leans towards small - cap growth + high - quality. The factor exposure and historical performance are presented, and the portfolio's monthly returns and August holdings are also provided [25][30].

Quantitative Models and Construction Methods 1. Model Name: Volume Timing Signal - **Model Construction Idea**: The core idea is that "the decline in a shrinking volume market is significantly greater than the rise in a shrinking volume market, so avoiding shrinking volume signals can achieve higher trading odds"[3][22][24] - **Model Construction Process**: 1. Calculate the rolling 60-day average and standard deviation of the turnover and turnover rate of the index or market[23] 2. Standardize the daily turnover data: - If the turnover is within ±2 standard deviations, map the score to -1~+1 - If the turnover exceeds ±2 standard deviations, assign a score of +1/-1 3. Combine the scores of turnover and turnover rate equally[23] 4. Generate signals based on the combined score: - Method 1: Go long if the score > 0, stay out if the score < 0 - Method 2: Use the rolling 5-year or 3-year percentile of the score; go long if above the 50th percentile, stay out if below[23] 5. The report adopts the simpler method of directly judging whether the score is greater than 0[23] - **Model Evaluation**: The model is not a high-win-rate strategy but achieves relatively high odds by avoiding significant market adjustments during shrinking volume periods[24] 2. Model Name: Growth-Value Style Rotation Model - **Model Construction Idea**: The model evaluates the relative attractiveness of growth and value styles based on macroeconomic cycles, valuation differences, and market sentiment[52][54] - **Model Construction Process**: 1. **Fundamentals**: - Growth is favored when the profit cycle slope is steep, interest rate levels are low, and the credit cycle is rising - Value is favored under the opposite conditions[52] 2. **Valuation**: - Growth is favored when the PE and PB valuation differences between growth and value are in the lower percentiles and mean-reverting upward[52] 3. **Sentiment**: - Growth is favored when turnover and volatility differences between growth and value are low[52] 4. Combine signals from fundamentals, valuation, and sentiment to determine the allocation between growth and value[52] - **Model Evaluation**: The model has shown significant improvement over the benchmark in terms of annualized returns and risk-adjusted performance[53][55] 3. Model Name: Small-Cap vs. Large-Cap Style Rotation Model - **Model Construction Idea**: The model evaluates the relative attractiveness of small-cap and large-cap styles based on macroeconomic cycles, valuation differences, and market sentiment[56][58] - **Model Construction Process**: 1. **Fundamentals**: - Small-cap is favored when the profit cycle slope is steep, interest rate levels are low, and the credit cycle is rising - Large-cap is favored under the opposite conditions[56] 2. **Valuation**: - Large-cap is favored when the PE and PB valuation differences between small-cap and large-cap are in the higher percentiles and mean-reverting downward[56] 3. **Sentiment**: - Small-cap is favored when turnover differences are high - Large-cap is favored when volatility differences are mean-reverting downward[56] 4. Combine signals from fundamentals, valuation, and sentiment to determine the allocation between small-cap and large-cap[56] - **Model Evaluation**: The model has shown significant improvement over the benchmark in terms of annualized returns and risk-adjusted performance[57][60] 4. Model Name: Four-Style Rotation Model - **Model Construction Idea**: Combines the conclusions of the growth-value and small-cap-large-cap rotation models to allocate across four styles: small-cap growth, small-cap value, large-cap growth, and large-cap value[61][63] - **Model Construction Process**: 1. Use the growth-value model to determine the allocation between growth and value 2. Use the small-cap-large-cap model to determine the allocation between small-cap and large-cap 3. Combine the two models to allocate across the four styles[61] - **Model Evaluation**: The model has shown significant improvement over the benchmark in terms of annualized returns and risk-adjusted performance, with consistent outperformance in most years[61][63] --- Model Backtest Results 1. Volume Timing Signal - **Win Rate**: 47.34%[24] - **Odds**: 1.75[24] - **Annualized Excess Return**: 6.87% (based on next-day open price)[34] - **Maximum Drawdown**: 31.40%[34] - **Return-to-Drawdown Ratio**: 0.4634[34] 2. Growth-Value Style Rotation Model - **Annualized Return**: 11.76%[55] - **Annualized Volatility**: 20.77%[55] - **Maximum Drawdown**: 43.07%[55] - **Sharpe Ratio**: 0.5438[55] - **Return-to-Drawdown Ratio**: 0.2731[55] 3. Small-Cap vs. Large-Cap Style Rotation Model - **Annualized Return**: 12.45%[60] - **Annualized Volatility**: 22.65%[60] - **Maximum Drawdown**: 50.65%[60] - **Sharpe Ratio**: 0.5441[60] - **Return-to-Drawdown Ratio**: 0.2459[60] 4. Four-Style Rotation Model - **Annualized Return**: 13.37%[63] - **Annualized Volatility**: 21.51%[63] - **Maximum Drawdown**: 47.91%[63] - **Sharpe Ratio**: 0.5988[63] - **Return-to-Drawdown Ratio**: 0.2790[63]

Quantitative Models and Construction Methods - **Model Name**: Style Rotation Model **Model Construction Idea**: The model is built from micro-level stock characteristics, leveraging valuation, market capitalization, volatility, and momentum factors to construct a style timing and scoring system. It integrates micro-level indicators and machine learning techniques to optimize style rotation strategies[4][9] **Model Construction Process**: 1. Select 80 base factors as original features based on the Dongwu multi-factor system[9] 2. Construct 640 micro-level features from these base factors[4][9] 3. Replace absolute proportion division of style factors with common indices as style stock pools to create new style returns as labels[4][9] 4. Use rolling training with a Random Forest model to avoid overfitting risks, optimize feature selection, and generate style recommendations[4][9] 5. Develop a framework from style timing to scoring, and from scoring to actual investment decisions[9] **Model Evaluation**: The model effectively avoids overfitting risks and provides a comprehensive framework for style rotation strategies[9] Model Backtesting Results - **Style Rotation Model**: - Annualized Return: 16.66%[10][11] - Annualized Volatility: 19.57%[10][11] - Information Ratio (IR): 0.85[10][11] - Monthly Win Rate: 56.31%[10][11] - Maximum Drawdown: -29.34%[11] - Excess Return (vs Benchmark): 11.40%[10][11] - Excess Volatility (vs Benchmark): 13.04%[10][11] - Excess IR (vs Benchmark): 0.87[10][11] - Excess Monthly Win Rate (vs Benchmark): 57.28%[10][11] - Excess Maximum Drawdown (vs Benchmark): -9.73%[11] Quantitative Factors and Construction Methods - **Factor Name**: Valuation, Market Capitalization, Volatility, Momentum **Factor Construction Idea**: These factors are derived from micro-level stock characteristics and are used to construct style timing and scoring systems[4][9] **Factor Construction Process**: 1. Extract micro-level features from base factors[4][9] 2. Use these features to create style returns as labels for machine learning models[4][9] 3. Apply Random Forest models to optimize factor selection and timing[4][9] **Factor Evaluation**: These factors are foundational to the style rotation model and contribute to its effectiveness in timing and scoring[4][9] Factor Backtesting Results - **Valuation Factor**: Monthly Returns (2025/01-2025/05): -2.00%, 0.00%, 2.00%, 4.00%, 6.00%[13][20] - **Market Capitalization Factor**: Monthly Returns (2025/01-2025/05): -4.00%, -2.00%, 0.00%, 2.00%, 4.00%[13][20] - **Volatility Factor**: Monthly Returns (2025/01-2025/05): -6.00%, -4.00%, -2.00%, 0.00%, 2.00%[13][20] - **Momentum Factor**: Monthly Returns (2025/01-2025/05): -8.00%, -6.00%, -4.00%, -2.00%, 0.00%[13][20]

Quantitative Models and Construction Methods 1. Model Name: Short-term Timing Model - **Model Construction Idea**: The model aims to provide short-term market timing signals based on various market indicators. - **Model Construction Process**: - **Fundamental Indicators**: - Manufacturing PMI: Current value is 49.70, at the 44.92% percentile over the past 5 years, giving a neutral signal[17] - RMB medium and long-term loan balance growth rate: Current value is 6.78%, at the 0.00% percentile over the past 5 years, giving a cautious signal[17] - M1 growth rate: Current value is 2.30%, at the 77.97% percentile over the past 5 years, giving an optimistic signal[17] - **Valuation Indicators**: - PE median: Current value is 40.16, at the 92.80% percentile over the past 5 years, giving a neutral signal[18] - PB median: Current value is 2.68, at the 71.05% percentile over the past 5 years, giving a neutral signal[18] - **Sentiment Indicators**: - Beta dispersion: Current value is -0.59%, at the 40.68% percentile over the past 5 years, giving a neutral signal[20] - Volume sentiment score: Current value is 0.30, at the 72.70% percentile over the past 5 years, giving an optimistic signal[20] - Volatility: Current value is 11.57% (annualized), at the 12.99% percentile over the past 5 years, giving a neutral signal[20] - **Liquidity Indicators**: - Monetary rate indicator: Current value is -0.10, at the 33.90% percentile over the past 5 years, giving an optimistic signal[20] - Exchange rate expectation indicator: Current value is -0.09%, at the 40.68% percentile over the past 5 years, giving a neutral signal[20] - Average new financing amount over 5 days: Current value is 23.20 billion, at the 80.81% percentile over the past 5 years, giving a neutral signal[20] - **Model Evaluation**: The model provides a comprehensive view of short-term market conditions by integrating fundamental, valuation, sentiment, and liquidity indicators. 2. Model Name: Growth-Value Style Rotation Model - **Model Construction Idea**: The model aims to rotate between growth and value styles based on economic cycles and market conditions. - **Model Construction Process**: - **Fundamental Indicators**: - Profit cycle slope: High, favoring growth[32] - Interest rate cycle level: High, favoring value[32] - Credit cycle trend: Weak, favoring value[32] - **Valuation Indicators**: - PE valuation difference: 5-year percentile is 15.19%, favoring growth[32] - PB valuation difference: 5-year percentile is 34.08%, favoring growth[32] - **Sentiment Indicators**: - Turnover difference: 5-year percentile is 21.01%, favoring value[32] - Volatility difference: 5-year percentile is 20.58%, favoring balanced allocation[32] - **Model Evaluation**: The model effectively captures the rotation between growth and value styles by considering fundamental, valuation, and sentiment factors. 3. Model Name: Small-Cap vs. Large-Cap Style Rotation Model - **Model Construction Idea**: The model aims to rotate between small-cap and large-cap styles based on economic cycles and market conditions. - **Model Construction Process**: - **Fundamental Indicators**: - Profit cycle slope: High, favoring small-cap[36] - Interest rate cycle level: High, favoring large-cap[36] - Credit cycle trend: Weak, favoring large-cap[36] - **Valuation Indicators**: - PE valuation difference: 5-year percentile is 80.60%, favoring large-cap[36] - PB valuation difference: 5-year percentile is 99.59%, favoring large-cap[36] - **Sentiment Indicators**: - Turnover difference: 5-year percentile is 54.26%, neutral[36] - Volatility difference: 5-year percentile is 83.71%, favoring large-cap[36] - **Model Evaluation**: The model provides a balanced approach to rotating between small-cap and large-cap styles by integrating fundamental, valuation, and sentiment indicators. 4. Model Name: Four-Style Rotation Model - **Model Construction Idea**: The model combines the growth-value and small-cap vs. large-cap rotation models to provide a comprehensive allocation across four styles. - **Model Construction Process**: - **Allocation Recommendation**: - Small-cap growth: 12.5%[41] - Small-cap value: 37.5%[41] - Large-cap growth: 12.5%[41] - Large-cap value: 37.5%[41] - **Model Evaluation**: The model offers a diversified approach to style rotation, leveraging insights from both growth-value and small-cap vs. large-cap models. Model Backtest Results Short-term Timing Model - Annualized Return: 16.58%[26] - Annualized Volatility: 14.57%[26] - Maximum Drawdown: 27.70%[26] - Sharpe Ratio: 0.9889[26] - Monthly Win Rate: 69.74%[26] - Quarterly Win Rate: 69.23%[26] - Annual Win Rate: 85.71%[26] Growth-Value Style Rotation Model - Annualized Return: 11.67%[35] - Annualized Volatility: 20.84%[35] - Maximum Drawdown: 43.07%[35] - Sharpe Ratio: 0.5387[35] - Monthly Win Rate: 58.28%[35] - Quarterly Win Rate: 60.78%[35] Small-Cap vs. Large-Cap Style Rotation Model - Annualized Return: 12.21%[40] - Annualized Volatility: 22.73%[40] - Maximum Drawdown: 50.65%[40] - Sharpe Ratio: 0.5336[40] - Monthly Win Rate: 60.93%[40] - Quarterly Win Rate: 58.82%[40] Four-Style Rotation Model - Annualized Return: 13.17%[43] - Annualized Volatility: 21.58%[43] - Maximum Drawdown: 47.91%[43] - Sharpe Ratio: 0.5895[43] - Monthly Win Rate: 59.60%[43] - Quarterly Win Rate: 62.75%[43] - Annual Win Rate: 69.23%[43]

- Model Name: Style Rotation Model; Model Construction Idea: The model is built from basic style factors such as valuation, market capitalization, volatility, and momentum, gradually constructing a style timing and scoring system[1][6] - Model Construction Process: 1. Construct 640 micro features based on 80 underlying micro indicators[1][6] 2. Use common indices as style stock pools instead of absolute proportion division of style factors to construct new style returns as labels[1][6] 3. Use a rolling training random forest model to avoid overfitting risks, select features, and obtain style recommendations[1][6] 4. Construct a style rotation framework from style timing to style scoring and from style scoring to actual investment[1][6] - Model Evaluation: The model effectively avoids overfitting risks and provides a comprehensive framework for style rotation from timing to scoring and actual investment[1][6] Model Backtest Results - Style Rotation Model, Annualized Return: 21.63%, Annualized Volatility: 24.09%, IR: 0.90, Monthly Win Rate: 59.12%, Maximum Drawdown: 28.33%[7][8] - Market Benchmark, Annualized Return: 7.21%, Annualized Volatility: 21.56%, IR: 0.33, Monthly Win Rate: 56.20%, Maximum Drawdown: 43.34%[8] - Excess Return, Annualized Return: 13.35%, Annualized Volatility: 11.43%, IR: 1.17, Monthly Win Rate: 66.42%, Maximum Drawdown: 10.28%[7][8] Monthly Performance - June 2025, Style Rotation Model Return: 1.28%, Excess Return: -2.51%[13] - July 2025, Latest Style Timing Directions: Low Valuation, Small Market Cap, Reversal, Low Volatility[13] - July 2025, Latest Holding Index: CSI Dividend Index[13]

Group 1 - The core viewpoint of the article indicates that the style rotation model has been continuously validated, with macroeconomic factors driving large-cap and value signals in Q2 2025. In May, the market favored large-cap and value styles, with large-cap outperforming small-cap by 0.56% and value outperforming growth by 3.40% [1][2]. - In May, the single-factor multi-strategy model showed a monthly return of 3.31%, with an excess return of 0.33% relative to the benchmark [1][2]. - The dual-driven rotation strategy for large-cap and value signals received a composite score of 3, predicting a favorable outlook for large-cap and value styles in Q2 2025 [1][2]. Group 2 - The industry rotation model for May indicated that the single-factor multi-strategy outperformed with an excess return of 0.33%, while the composite factor strategy had an excess return of -0.64% [2]. - For June, the recommended long positions in the single-factor multi-strategy include non-bank financials, electronics, and banks, while the composite factor strategy recommends non-bank financials, pharmaceuticals, building materials, basic chemicals, and steel [2].