证券研究报告 金工 节前增配大盘价值，成长内高低切 2025 年 9 月 28 日│中国内地 量化投资周报 本周观点：节前增配大盘价值，成长风格内高低切 从上周盘面来看，虽然指数箱体震荡，且大盘择时模型继续看多，但整体缩 量，表明高位板块的后续接力难度加大，且市场情绪出现明显波动——周二 全天和周三开盘红利风格显著占优，周三盘中在云栖大会利好消息催化下成 长风格重拾升势，但周五再次回到红利风格占优的局面，红利风格或已出现 止跌迹象。此外，10 年期美债利率连续两周上行合计达 10 bp，且我们预测 9 月美国 CPI 同比增速或回升至 3.2%左右——10 月美联储能否如期降息仍 然存疑。流动性环境边际收紧。结合风格择时和行业轮动模型，叠加投资者 对长假不确定性的担忧，建议节前增配大盘价值、成长风格内高低切。 A 股大盘择时模型：继续看多 A 股大盘 我们以万得全 A 指数作为 A 股大盘代理，从估值、情绪、资金、技术四个 维度对 A 股大盘进行整体方向性判断。今年以来，模型多空择时的扣费后 收益 40.98%，同期 A 股大盘涨跌幅为 23.96%，超额收益为 17.02%；上周 模型超额收益为-0.10% ...

Group 1 - The report focuses on constructing a style timing and rotation model from a micro perspective, utilizing micro data to enhance the strategy system [6][62] - The model is based on four style factors: valuation, market capitalization, volatility, and momentum, using 80 micro indicators to create a scoring system [7][62] - The backtesting period from January 1, 2014, to July 31, 2025, shows an annualized return of 20.90% with a volatility of 26.12% and a maximum drawdown of -40.57% [57][62] Group 2 - The model's out-of-sample performance has been stable since its development in March 2024, with a return of 55.36% for the entire year of 2024, outperforming the market benchmark by 35.72% [57][62] - The report highlights the construction of style labels based on specific broad indices to overcome limitations of using the entire A-share market for style timing [18][20] - The random forest model is selected for predicting the direction of style factors, enhancing the performance of the timing strategy [23][25] Group 3 - The performance metrics for the valuation factor before timing show an annualized return of 7.90% compared to the benchmark's 6.85%, with a maximum drawdown of -60.33% [30][32] - After applying the timing model, the valuation factor's annualized return improves to 15.19%, significantly outperforming the benchmark [38][40] - The momentum factor shows a pre-timing annualized return of 10.11%, which increases to 15.73% post-timing, indicating improved performance [42][47] Group 4 - The volatility factor's pre-timing performance indicates an annualized return of 10.93%, while post-timing performance shows an increase to 15.73% [48][53] - The equal-weighted composite factor, derived from the four style factors, achieves an annualized return of 20.05% with a maximum drawdown of -41.97% [52][55] - The scoring system for the style factors is based on historical prediction accuracy, further refining the composite factor's performance [56][59]

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]

- 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]

Core Viewpoint - The investment strategy of the company has evolved from a singular focus on growth stocks to a diversified approach that adapts to market changes, emphasizing the importance of both sustainable growth and risk management in investment decisions [1][5][9]. Group 1: Investment Strategy Evolution - The company has recognized the limitations of a single investment strategy, especially in the current challenging market for growth stocks, prompting a shift towards diversification [1][4]. - The investment philosophy now incorporates a combination of growth, value, and dividend stocks, with a focus on macroeconomic trends and style timing to enhance portfolio resilience [5][9]. - The company aims to balance investment opportunities with safety, particularly in sectors benefiting from AI advancements and those with reasonable valuations [1][9]. Group 2: Market Outlook and Focus Areas - The company believes that the most uncertain phase of the market has passed, with expectations for better investment opportunities in the fourth quarter, particularly in growth stocks [9]. - Key sectors of interest include the internet sector benefiting from AI development, domestic consumption-related industries, technology with a focus on self-sufficiency, and sectors supported by growth policies like engineering machinery [9][10]. - The semiconductor industry is highlighted for its significant growth potential, driven by increasing domestic production and technological advancements [10].

Quantitative Models and Construction Methods - **Model Name**: Style Rotation Model **Model Construction Idea**: The model is built from micro-level stock factors, focusing on valuation, market capitalization, volatility, and momentum. It integrates a style timing and scoring system to construct a monthly frequency style rotation framework[1][6] **Model Construction Process**: 1. Start with 80 base micro-level factors selected based on the Dongwu multi-factor system[6] 2. Generate 640 micro-level features from these base factors[6] 3. Replace the absolute proportion division of style factors with commonly used indices as style stock pools to create new style returns as labels[6] 4. Use a rolling training random forest model to avoid overfitting risks, optimize feature selection, and derive style recommendations[6] 5. Construct a framework that transitions from style timing to style scoring and finally to actual investment decisions[6] **Model Evaluation**: The model effectively avoids overfitting risks and provides a comprehensive framework for style rotation[6] Model Backtesting Results - **Style Rotation Model**: - Annualized Return: 21.56% - Annualized Volatility: 24.17% - IR: 0.89 - Monthly Win Rate: 58.82% - Maximum Drawdown: 28.33%[7][8] - Excess Performance (Hedged Against Benchmark): - Annualized Return: 13.45% - Annualized Volatility: 11.47% - IR: 1.17 - Monthly Win Rate: 66.18% - Maximum Drawdown: 10.28%[7][8] Quantitative Factors and Construction Methods - **Factor Name**: Valuation, Market Capitalization, Volatility, Momentum **Factor Construction Idea**: These are foundational style factors used to construct the style rotation model. They are further refined into micro-level features and integrated into the model's scoring and timing system[1][6] **Factor Construction Process**: 1. Valuation: Derived from traditional valuation metrics such as P/E, P/B, and dividend yield[6] 2. Market Capitalization: Categorized into large-cap and small-cap stocks based on market size[6] 3. Volatility: Measured using historical price fluctuations[6] 4. Momentum: Calculated based on past price trends and returns[6] Factor Backtesting Results - **Factor Performance (2025, Multi-Factor Timing Results)**: - Valuation: -2.00% - Market Capitalization: 4.00% - Volatility: -6.00% - Momentum: -8.00%[10][17] - **Factor Performance (2025, Actual Factor Returns)**: - Valuation: 2.00% - Market Capitalization: 6.00% - Volatility: -4.00% - Momentum: -8.00%[10][11] Additional Notes - **Latest Style Timing Directions (May 2025)**: Value, Large-Cap, Reversal, Low Volatility[14] - **Latest Holding Index (May 2025)**: CSI Dividend Index[15]

Core Viewpoint - The article emphasizes the importance of a multi-strategy dynamic allocation approach to capture style rotation opportunities in the market, utilizing quantitative indicators to assess the allocation value of different styles or strategies [1][6]. Summary by Sections Style Timing Framework to Strategy Rotation Model - The style timing model can effectively avoid high-risk phases but may miss some upward opportunities in styles. Historical data is used to identify similar past indicators to predict future performance [3][26]. - A voting method is employed to integrate multiple indicators, resulting in a comprehensive style timing model that has shown to reduce risk while maintaining a lower annualized return compared to holding styles directly [3][31]. Performance Metrics - The style timing model achieved an annualized return of 16.5% during the backtest period from January 1, 2015, to January 31, 2025, with an excess return of 12.7% over the benchmark [3][39]. - The active quantitative strategy rotation model yielded an annualized return of 36.2% during the backtest period from January 1, 2015, to February 28, 2025, outperforming the benchmark by 28.5% [4][39]. Key Indicators for Style Allocation - The article identifies key indicators for measuring style allocation value, including valuation difference, active inflow rate difference, and combination temporal correlation [2][17]. - Historical data shows that a larger valuation difference correlates with better future excess returns, while a significant active inflow rate difference indicates potential overreaction risks [2][10]. Latest Insights and Recommendations - As of March 2025, the recommendation is to favor small-cap and growth styles while maintaining a neutral stance on value and dividend styles [4][35]. - The report suggests holding indices like the CSI 2000 for small-cap and the National Growth Index for growth styles, along with specific active quantitative strategies [4][35]. Multi-Dimensional Timing Indicators - The article discusses the construction of a multi-dimensional timing indicator system that includes valuation difference, market participation, and combination consistency to assess future style performance [18][22]. - The effectiveness of these indicators is tested, showing that they can provide valuable insights into future excess returns across different styles [22][23]. Strategy Rotation and Dynamic Allocation - The article outlines a strategy for dynamic allocation and rotation among styles based on multi-dimensional timing indicators, aiming to optimize returns while managing risks [37][39]. - The dynamic allocation strategy is designed to adjust holdings based on the prevailing market conditions and style performance indicators [37][39].