量化大势研判 202601 宜攻守兼备：成长+质量 glmszqdatemark [Table_Author] 相关研究 本公司具备证券投资咨询业务资格，请务必阅读最后一页免责声明 证券研究报告 1 2026 年 01 月 04 日 分析师 叶尔乐 执业证书： S0590525110059 邮箱： yeerle@glms.com.cn 为了解决风格系统化轮动难题，我们开启了自下而上的量化大势研判探究。所谓 风格是资产本身的内在属性，内嵌于其产业生命周期的变化中，存在五种风格阶 段：外延成长、质量成长、质量红利、价值红利、破产价值。全局比较下来的优 势资产其特征属性，即为未来市场的主流风格。可通过 g>ROE>D 的基本优先级 进行考察，分析"有没有（好资产）"和"（好资产）贵不贵"进行优势资产的比 较判断，最后聚焦当下最具有优势的细分板块。 预期增速资产优势差继续回升，实际增速资产优势差小幅回升，继续推荐预期成 长风格，景气类成长建议标配。△gf 继续保持扩张，其主要来自 Top 组的上升， 高预期增速板块依然有表现机会；△g 小幅回升，景气类策略虽缺乏基本面动量优 势，但可保持标配。 ROE 的优势差继续回 ...

Quantitative Models and Construction Methods 1. Model Name: Quantitative Market Trend Judgment Framework - **Model Construction Idea**: The model aims to address the systematic rotation of market styles by identifying the dominant asset characteristics that represent the future mainstream market style. It evaluates assets based on the priority of "g > ROE > D" to determine whether there are good assets and whether they are overvalued[5][8][12] - **Model Construction Process**: 1. Define five style stages based on the industry lifecycle: external growth, quality growth, quality dividend, value dividend, and bankruptcy value[8] 2. Use the "g > ROE > D" priority to compare assets, focusing on growth (g), profitability (ROE), and dividend yield (D)[5][8] 3. Incorporate factors such as expected growth (gf), actual growth (g), profitability (ROE), and valuation metrics (PB, DP, BP) to classify and evaluate assets[9][12] 4. Apply the framework to select industries and allocate them equally within each strategy[19] - **Model Evaluation**: The framework has demonstrated strong explanatory power for A-share market style rotation since 2009, achieving an annualized return of 27.06%[19] --- Model Backtesting Results Quantitative Market Trend Judgment Framework - **Annualized Return**: 27.06% since 2009[19] - **Excess Returns by Year**: - 2017: 27% - 2020: 44% - 2022: 62% - 2024: 52% - 2025 (YTD): 8%[22] --- Quantitative Factors and Construction Methods 1. Factor Name: Expected Growth (gf) - **Factor Construction Idea**: Measures the expected growth rate of industries based on analysts' forecasts, regardless of the lifecycle stage[9] - **Factor Construction Process**: 1. Calculate the expected growth rate (gf) for each industry 2. Rank industries based on the highest expected growth rates 3. Select top-performing industries for allocation[9][38] - **Factor Evaluation**: The factor has shown significant excess returns since 2019, with notable performance in 2014-2015 and 2025[38] 2. Factor Name: Actual Growth (g) - **Factor Construction Idea**: Focuses on industries with the highest earnings momentum (△g), particularly during transition and growth phases[9] - **Factor Construction Process**: 1. Use △g to represent earnings momentum 2. Rank industries based on △g and select the top-performing ones 3. Incorporate additional factors such as SUE, SUR, and JOR for refinement[40] - **Factor Evaluation**: The factor has consistently delivered significant excess returns, especially in growth-dominant environments[40] 3. Factor Name: Profitability (ROE) - **Factor Construction Idea**: Targets industries with high ROE and low valuation under the PB-ROE framework, focusing on mature stages[9] - **Factor Construction Process**: 1. Calculate PB-ROE residuals for each industry 2. Rank industries based on residuals and select the top-performing ones[43] - **Factor Evaluation**: The factor performed strongly from 2016 to 2020 but has weakened since 2021[43] 4. Factor Name: Quality Dividend (DP + ROE) - **Factor Construction Idea**: Combines dividend yield (DP) and ROE to identify industries with the highest scores, focusing on mature stages[9] - **Factor Construction Process**: 1. Calculate DP and ROE for each industry 2. Combine the two metrics into a composite score 3. Rank industries and select the top-performing ones[46] - **Factor Evaluation**: The factor has shown significant excess returns in 2016, 2017, and 2023[46] 5. Factor Name: Value Dividend (DP + BP) - **Factor Construction Idea**: Combines dividend yield (DP) and book-to-price ratio (BP) to identify undervalued industries, focusing on mature stages[9] - **Factor Construction Process**: 1. Calculate DP and BP for each industry 2. Combine the two metrics into a composite score 3. Rank industries and select the top-performing ones[49] - **Factor Evaluation**: The factor has delivered significant excess returns in 2009, 2017, and 2021-2023[49] 6. Factor Name: Bankruptcy Value (PB + SIZE) - **Factor Construction Idea**: Targets industries with the lowest PB and SIZE scores, focusing on stagnation and recession stages[9] - **Factor Construction Process**: 1. Calculate PB and SIZE for each industry 2. Combine the two metrics into a composite score 3. Rank industries and select the lowest-scoring ones[52] - **Factor Evaluation**: The factor has shown significant excess returns in 2015-2016 and 2021-2023[52] --- Factor Backtesting Results Expected Growth (gf) - **Recent Performance**: - Lithium: +51.15% (3 months) - Frozen Food: +14.08% (3 months)[38] Actual Growth (g) - **Recent Performance**: - Lithium Chemicals: +51.88% (3 months) - Other Home Appliances: +14.93% (3 months)[41] Profitability (ROE) - **Recent Performance**: - Network Equipment: +12.18% (3 months) - Buses: +10.46% (3 months)[43] Quality Dividend (DP + ROE) - **Recent Performance**: - Timber Processing: +145.24% (3 months) - Lithium Equipment: +21.95% (3 months)[46] Value Dividend (DP + BP) - **Recent Performance**: - Network Equipment: +12.18% (3 months) - Security: -2.24% (3 months)[49] Bankruptcy Value (PB + SIZE) - **Recent Performance**: - Gas: +15.88% (3 months) - Building Renovation: +16.42% (3 months)[52]

Quantitative Models and Construction Methods Model Name: Quantitative Market Trend Analysis Framework - **Model Construction Idea**: The model aims to solve the systematic rotation problem of styles by conducting a bottom-up quantitative market trend analysis. It identifies the dominant asset characteristics that represent the future market's mainstream style through a comprehensive comparison of assets[1][5] - **Model Construction Process**: - The model considers five style stages based on the asset's industry lifecycle: external growth, quality growth, quality dividend, value dividend, and bankruptcy value[1][5] - The priority for asset comparison is based on the sequence: growth (g) > return on equity (ROE) > dividend (D)[1][5] - The model uses the spread of asset advantage differences to capture the trend changes of top assets, similar to factor timing[20] - **Model Evaluation**: The framework has shown good explanatory power for past A-share style rotations, achieving an annualized return of 27.25% since 2009[15] Model Backtesting Results - **Quantitative Market Trend Analysis Framework**: - 2009: Asset Comparison Strategy 133%, Wind All A 82%, Excess Return 51%[18] - 2010: Asset Comparison Strategy 7%, Wind All A -7%, Excess Return 14%[18] - 2011: Asset Comparison Strategy -33%, Wind All A -22%, Excess Return -11%[18] - 2012: Asset Comparison Strategy 5%, Wind All A 5%, Excess Return 0%[18] - 2013: Asset Comparison Strategy 41%, Wind All A 5%, Excess Return 36%[18] - 2014: Asset Comparison Strategy 48%, Wind All A 52%, Excess Return -4%[18] - 2015: Asset Comparison Strategy 55%, Wind All A 38%, Excess Return 16%[18] - 2016: Asset Comparison Strategy -14%, Wind All A -13%, Excess Return -1%[18] - 2017: Asset Comparison Strategy 32%, Wind All A 5%, Excess Return 27%[18] - 2018: Asset Comparison Strategy -21%, Wind All A -28%, Excess Return 7%[18] - 2019: Asset Comparison Strategy 41%, Wind All A 33%, Excess Return 8%[18] - 2020: Asset Comparison Strategy 69%, Wind All A 26%, Excess Return 44%[18] - 2021: Asset Comparison Strategy 47%, Wind All A 9%, Excess Return 38%[18] - 2022: Asset Comparison Strategy 44%, Wind All A -19%, Excess Return 62%[18] - 2023: Asset Comparison Strategy 5%, Wind All A -5%, Excess Return 10%[18] - 2024: Asset Comparison Strategy 62%, Wind All A 10%, Excess Return 52%[18] - 2025 (Aug): Asset Comparison Strategy 27%, Wind All A 23%, Excess Return 4%[18] Quantitative Factors and Construction Methods Factor Name: Expected Growth (gf) - **Factor Construction Idea**: The factor focuses on the highest analyst forecasted growth rates, regardless of the cycle stage[6] - **Factor Construction Process**: - The factor is constructed by selecting industries with the highest expected growth rates as forecasted by analysts[6] - The spread of expected growth advantage differences (Δgf) is used to capture the trend changes in top assets[20] - **Factor Evaluation**: The factor has shown significant excess returns since 2019, with notable performance in 2014-2015[34] Factor Name: Actual Growth (g) - **Factor Construction Idea**: The factor focuses on industries with the highest actual growth rates, particularly during transition and growth periods[6] - **Factor Construction Process**: - The factor is constructed by selecting industries with the highest actual growth rates (Δg)[6] - The spread of actual growth advantage differences (Δg) is used to capture the trend changes in top assets[24] - **Factor Evaluation**: The factor has shown significant excess returns in growth-dominant environments[36] Factor Name: Profitability (ROE) - **Factor Construction Idea**: The factor focuses on industries with high ROE and low valuation under the PB-ROE framework, concentrated in mature periods[6] - **Factor Construction Process**: - The factor is constructed by selecting industries with high ROE and low PB-ROE residuals[6] - The spread of ROE advantage differences is used to capture the trend changes in top assets[26] - **Factor Evaluation**: The factor has shown significant excess returns from 2016 to 2020, with weaker performance since 2021[39] Factor Name: Quality Dividend (DP+ROE) - **Factor Construction Idea**: The factor focuses on industries with the highest DP+ROE scores, concentrated in mature periods[6] - **Factor Construction Process**: - The factor is constructed by selecting industries with the highest DP+ROE scores[6] - The spread of DP+ROE advantage differences is used to capture the trend changes in top assets[42] - **Factor Evaluation**: The factor has shown significant excess returns in 2016, 2017, and 2023[43] Factor Name: Value Dividend (DP+BP) - **Factor Construction Idea**: The factor focuses on industries with the highest DP+BP scores, concentrated in mature periods[6] - **Factor Construction Process**: - The factor is constructed by selecting industries with the highest DP+BP scores[6] - The spread of DP+BP advantage differences is used to capture the trend changes in top assets[45] - **Factor Evaluation**: The factor has shown significant excess returns in 2009, 2017, and 2021-2023[46] Factor Name: Bankruptcy Value (PB+SIZE) - **Factor Construction Idea**: The factor focuses on industries with the lowest PB+SIZE scores, concentrated in stagnation and recession periods[6] - **Factor Construction Process**: - The factor is constructed by selecting industries with the lowest PB+SIZE scores[6] - The spread of PB+SIZE advantage differences is used to capture the trend changes in top assets[48] - **Factor Evaluation**: The factor has shown significant excess returns in 2015-2016 and 2021-2023[49] Factor Backtesting Results - **Expected Growth (gf)**: - Cable: 12 stocks, largest weight stock Zhongtian Technology, average market cap 21.791 billion yuan, 3-month performance 49.62%[34] - Cement: 19 stocks, largest weight stock Conch Cement, average market cap 17.929 billion yuan, 3-month performance 12.71%[34] - Glass Fiber: 6 stocks, largest weight stock China Jushi, average market cap 26.657 billion yuan, 3-month performance 63.67%[34] - Rare Earth and Magnetic Materials: 17 stocks, largest weight stock Northern Rare Earth, average market cap 31.018 billion yuan, 3-month performance 98.77%[34] - White Goods III: 10 stocks, largest weight stock Midea Group, average market cap 113.675 billion yuan, 3-month performance -1.21%[34] - **Actual Growth (g)**: - Integrated Circuits: 104 stocks, largest weight stock Cambricon-U, average market cap 45.058 billion yuan, 3-month performance 42.93%[37] - PCB: 38 stocks, largest weight stock Shenghong Technology, average market cap 27.163 billion yuan, 3-month performance 112.10%[37] - Tungsten: 4 stocks, largest weight stock Xiamen Tungsten, average market cap 30.523 billion yuan, 3-month performance 69.26%[37] - Lithium Battery Equipment: 12 stocks, largest weight stock Lead Intelligent, average market cap 11.731 billion yuan, 3-month performance 60.15%[37] - Weapons and Equipment III: 12 stocks, largest weight stock Great Wall Military Industry, average market cap 21.307 billion yuan, 3-month performance 80.22%[37] - **Profitability (ROE)**: - Beer: 7 stocks, largest weight stock Tsingtao Brewery, average market cap 26.758 billion yuan, 3-month performance -3.94%[39] - Liquor: 20 stocks, largest weight stock Kweichow Moutai, average market cap 162.722 billion yuan, 3-month performance 4.12%[39] - Non-dairy Beverages: 7 stocks, largest weight stock Eastroc Beverage, average market cap 32.754 billion yuan, 3-month performance -4.45%[39] - Network Connection and Tower Setup: 19 stocks, largest weight stock Zhongji Xuchuang, average market cap 64.299 billion yuan, 3-month performance 202.29%[39] - Building Decoration III: 28 stocks, largest weight stock Gold Mantis, average market cap 3.436 billion yuan, 3-month performance 4.42%[39] - **Quality Dividend (DP+ROE)**: - Automotive Motor Control: 15

Quantitative Models and Construction Methods 1. Model Name: Quantitative Market Trend Judgment Framework - **Model Construction Idea**: The framework identifies the dominant market style by comparing the intrinsic attributes of assets, which are tied to their industry lifecycle stages. It prioritizes assets based on the sequence of growth rate (g) > return on equity (ROE) > dividend yield (D) to determine the most advantageous assets and focuses on the most promising sectors[5][6][9] - **Model Construction Process**: 1. Define five style stages for equity assets: external growth, quality growth, quality dividend, value dividend, and distressed value[5] 2. Compare assets globally to identify advantageous ones based on their intrinsic characteristics[5] 3. Use the priority sequence g > ROE > D to evaluate whether good assets exist and whether they are overvalued[5][6] 4. Focus on sectors with the most advantageous characteristics in the current market[5][6] - **Model Evaluation**: The framework has demonstrated strong explanatory power for A-share market style rotations since 2009, achieving an annualized return of 26.70%[16] 2. Model Name: Asset Comparison Strategy - **Model Construction Idea**: This model categorizes assets into primary and secondary groups. Primary assets include actual growth, expected growth, and profitability assets. Secondary assets are prioritized based on crowding levels and fundamental factors[9] - **Model Construction Process**: 1. Classify assets into primary (expected growth, actual growth, profitability) and secondary (quality dividend, value dividend, distressed value) categories[9] 2. Allocate market funds to primary assets when any of them show an advantage; otherwise, shift to secondary assets[9] 3. Rank secondary assets by crowding levels and fundamental factors, with the order: quality dividend > value dividend > distressed value[9] --- Model Backtesting Results 1. Quantitative Market Trend Judgment Framework - Annualized return: 26.70% since 2009[16] - Historical performance: Positive excess returns in most years, with limited effectiveness in 2011, 2012, 2014, and 2016[16][19] - Excess returns by year: - 2009: 51% - 2010: 14% - 2013: 36% - 2017: 27% - 2020: 44% - 2022: 62%[19] --- Quantitative Factors and Construction Methods 1. Factor Name: Expected Growth (gf) - **Factor Construction Idea**: Measures the expected growth rate based on analysts' forecasts, regardless of the industry lifecycle stage[6] - **Factor Construction Process**: 1. Use analysts' forecasted growth rates as the primary input[6] 2. Calculate the spread (Δgf) between top and bottom groups to assess the trend of expected growth[21] - **Factor Evaluation**: The factor has shown consistent expansion, with top groups driving the increase, indicating analysts' optimism about high-growth sectors[21] 2. Factor Name: Actual Growth (g) - **Factor Construction Idea**: Focuses on performance momentum (Δg) during transition and growth phases[6] - **Factor Construction Process**: 1. Calculate the spread (Δg) between top and bottom groups based on actual growth rates[25] 2. Monitor the trend of Δg to identify growth opportunities in the market[25] - **Factor Evaluation**: The factor has shown gradual expansion, with opportunities in sectors maintaining strong momentum despite a slowdown in top-tier growth[25] 3. Factor Name: Profitability (ROE) - **Factor Construction Idea**: Evaluates valuation levels using the PB-ROE framework, focusing on mature industries[6] - **Factor Construction Process**: 1. Calculate the PB-ROE residuals for each industry[40] 2. Rank industries based on residuals to identify undervalued high-ROE sectors[40] - **Factor Evaluation**: The factor's advantage has declined, and its crowding level remains low, suggesting limited opportunities in the current market[28] 4. Factor Name: Quality Dividend (DP+ROE) - **Factor Construction Idea**: Combines dividend yield (DP) and ROE to identify high-quality dividend-paying industries[6] - **Factor Construction Process**: 1. Calculate DP and ROE scores for each industry[43] 2. Combine the scores to rank industries and select the top-performing ones[43] - **Factor Evaluation**: The factor has shown significant excess returns in specific years, such as 2016, 2017, and 2023[43] 5. Factor Name: Value Dividend (DP+BP) - **Factor Construction Idea**: Combines dividend yield (DP) and book-to-price ratio (BP) to identify undervalued dividend-paying industries[6] - **Factor Construction Process**: 1. Calculate DP and BP scores for each industry[47] 2. Combine the scores to rank industries and select the top-performing ones[47] - **Factor Evaluation**: The factor has demonstrated strong excess returns in years like 2009, 2017, and 2021-2023[47] 6. Factor Name: Distressed Value (PB+SIZE) - **Factor Construction Idea**: Identifies industries with low price-to-book ratios (PB) and small market capitalization (SIZE), focusing on stagnation and recession phases[6] - **Factor Construction Process**: 1. Calculate PB and SIZE scores for each industry[51] 2. Combine the scores to rank industries and select the lowest-scoring ones[51] - **Factor Evaluation**: The factor has shown significant excess returns during periods like 2015-2016 and 2021-2023[51] --- Factor Backtesting Results 1. Expected Growth (gf) - Δgf continues to expand, driven by top-tier groups, indicating analysts' optimism about high-growth sectors[21] 2. Actual Growth (g) - Δg shows gradual expansion, with opportunities in sectors maintaining strong momentum despite a slowdown in top-tier growth[25] 3. Profitability (ROE) - ROE advantage continues to decline, with low crowding levels and limited opportunities in the current market[28] 4. Quality Dividend (DP+ROE) - Significant excess returns in 2016, 2017, and 2023[43] 5. Value Dividend (DP+BP) - Strong excess returns in 2009, 2017, and 2021-2023[47] 6. Distressed Value (PB+SIZE) - Significant excess returns during 2015-2016 and 2021-2023[51]

Quantitative Models and Construction Methods 1. Model Name: Quantitative Market Trend Judgment Framework - **Model Construction Idea**: The framework identifies the dominant market style by comparing asset characteristics based on their intrinsic attributes and industry life cycle stages. It evaluates assets using the priority sequence of growth (g) > return on equity (ROE) > dividend yield (D) to determine the most advantageous sectors[5][6][7] - **Model Construction Process**: - Classifies assets into five style stages: external growth, quality growth, quality dividend, value dividend, and distressed value[5] - Prioritizes mainstream assets (actual growth, expected growth, profitability) over secondary assets, with secondary asset priority determined by crowding levels[7] - Uses metrics such as Δg (actual growth momentum), Δgf (expected growth momentum), PB-ROE (valuation), and DP (dividend yield) to evaluate and rank assets[6][7] - **Model Evaluation**: The framework has demonstrated strong explanatory power for A-share market style rotations since 2009, achieving an annualized return of 26.85%[14] --- Quantitative Factors and Construction Methods 1. Factor Name: Actual Growth (g) - **Factor Construction Idea**: Measures the momentum of actual growth by evaluating the change in net profit growth rates (Δg)[18] - **Factor Construction Process**: - Calculates the spread between the top and bottom quintiles of industries based on net profit growth rates[18] - Δg > 0 indicates strengthening growth momentum, while Δg < 0 signals weakening momentum[18] 2. Factor Name: Expected Growth (gf) - **Factor Construction Idea**: Captures analysts' optimism by assessing changes in expected net profit growth rates (Δgf)[21] - **Factor Construction Process**: - Similar to Δg, calculates the spread between the top and bottom quintiles of industries based on expected net profit growth rates[21] - Δgf > 0 indicates increasing optimism, while Δgf < 0 suggests declining optimism[21] 3. Factor Name: Profitability (ROE) - **Factor Construction Idea**: Evaluates asset quality by combining ROE with valuation metrics (PB-ROE)[22] - **Factor Construction Process**: - Ranks industries based on ROE and PB-ROE residuals[30] - Identifies high ROE industries with low valuations for allocation[30] 4. Factor Name: Quality Dividend (DP+ROE) - **Factor Construction Idea**: Combines dividend yield (DP) and ROE to identify high-quality dividend-paying industries[33] - **Factor Construction Process**: - Scores industries based on DP and ROE, selecting the highest-scoring sectors for allocation[33] 5. Factor Name: Value Dividend (DP+BP) - **Factor Construction Idea**: Combines dividend yield (DP) and book-to-price ratio (BP) to identify undervalued dividend-paying industries[36] - **Factor Construction Process**: - Scores industries based on DP and BP, selecting the highest-scoring sectors for allocation[36] 6. Factor Name: Distressed Value (PB+SIZE) - **Factor Construction Idea**: Identifies distressed industries with low price-to-book ratios (PB) and small market capitalization (SIZE)[38] - **Factor Construction Process**: - Scores industries based on PB and SIZE, selecting the lowest-scoring sectors for allocation[38] --- Backtesting Results of Models and Factors 1. Quantitative Market Trend Judgment Framework - **Annualized Return**: 26.85% since 2009[14] - **Excess Returns**: Positive in most years, with notable outperformance in 2020 (44%), 2021 (38%), and 2022 (62%)[15] 2. Actual Growth Factor - **Recent Performance**: Δg turned negative, indicating weakening growth momentum and reduced opportunities for actual growth strategies[18] 3. Expected Growth Factor - **Recent Performance**: Δgf turned negative, signaling a halt in analysts' optimism and limited opportunities for expected growth strategies[21] 4. Profitability Factor - **Recent Performance**: ROE advantage continues to decline, but reduced crowding levels suggest potential opportunities in a weakening growth environment[22][24] 5. Quality Dividend Factor - **Recent Performance**: Outperformed in 2016, 2017, and 2023; recent recommendations include sectors like "Other Home Appliances" and "Service Robots"[33] 6. Value Dividend Factor - **Recent Performance**: Outperformed in 2009, 2017, and 2021-2023; recent recommendations include sectors like "Pet Food" and "Security"[36] 7. Distressed Value Factor - **Recent Performance**: Outperformed in 2015-2016 and 2021-2023; recent recommendations include sectors like "Other Electronic Components" and "Printing and Dyeing"[38]