CMS(SH:600999) CMS·2025-08-03 11:05Quantitative Models and Construction Methods 1. Model Name: Credit Impulse Timing Strategy - Model Construction Idea: The model uses credit impulse as a timing indicator for A-shares, where the direction of credit impulse determines the market position (full position when upward, empty position when downward) [6][13][14] - Model Construction Process: - Calculate the year-on-year growth rate of long-term corporate loans (TTM) as the credit impulse indicator - Use the direction of the credit impulse to determine market positions: full position when the indicator is upward, empty position when downward - Formula: $ \text{Credit Impulse} = \frac{\text{Long-term Corporate Loans (TTM)} - \text{Long-term Corporate Loans (TTM, previous year)}}{\text{Long-term Corporate Loans (TTM, previous year)}} $ - Model Evaluation: The model has shown high effectiveness in avoiding major downtrends in the market [6][13][14] 2. Model Name: Beta Dispersion Timing Strategy - Model Construction Idea: The model uses beta dispersion as an indicator to measure local market sentiment overheating, with significant monthly timing effectiveness [6][17] - Model Construction Process: - Calculate the monthly beta dispersion of the market - Use the beta dispersion to determine market positions: higher beta dispersion indicates higher risk - Formula: $ \text{Beta Dispersion} = \frac{\sum_{i=1}^{N} (\beta_i - \bar{\beta})^2}{N} $ where $\beta_i$ is the beta of stock i, $\bar{\beta}$ is the average beta, and N is the number of stocks - Model Evaluation: The model has shown significant monthly timing effectiveness since 2013 [6][17] 3. Model Name: Trading Volume Timing Strategy - Model Construction Idea: The model uses trading volume as an indicator for market timing, with significant daily timing effectiveness [6][17] - Model Construction Process: - Calculate the daily trading volume and its 60-day moving average - Use the trading volume to determine market positions: higher trading volume indicates stronger market support - Formula: $ \text{Trading Volume Indicator} = \frac{\text{Daily Trading Volume}}{\text{60-day Moving Average of Trading Volume}} $ - Model Evaluation: The model has shown significant daily timing effectiveness since 2013 [6][17] 4. Composite Model: Credit Impulse, Beta Dispersion, Trading Volume - Model Construction Idea: The composite model combines credit impulse, beta dispersion, and trading volume indicators for market timing [6][18] - Model Construction Process: - Use equal weighting to combine the three indicators - Adjust positions based on the combined signal: average 2-week signal change frequency - Formula: $ \text{Composite Indicator} = \frac{\text{Credit Impulse Indicator} + \text{Beta Dispersion Indicator} + \text{Trading Volume Indicator}}{3} $ - Model Evaluation: The composite model has shown a high annual turnover rate and significant annualized returns since 2013 [6][18] Model Backtesting Results 1. Credit Impulse Timing Strategy - Annualized Return: 10.83% [6][13][14] - Avoided Major Downtrends: 2015 H2, 2018, 2022-2024 H1 [6][13][14] 2. Beta Dispersion Timing Strategy - Annualized Return: 13.12% [6][17] - Monthly Timing Effectiveness: Significant since 2013 [6][17] 3. Trading Volume Timing Strategy - Annualized Return: 14.33% [6][17] - Daily Timing Effectiveness: Significant since 2013 [6][17] 4. Composite Model: Credit Impulse, Beta Dispersion, Trading Volume - Annualized Return: 19.98% [6][18] - Annual Turnover Rate: 24 times [6][18] Quantitative Factors and Construction Methods 1. Factor Name: Manufacturing PMI Timing Strategy - Factor Construction Idea: The factor uses manufacturing PMI as a timing indicator for A-shares, with positions adjusted based on PMI levels [6][13] - Factor Construction Process: - Calculate the rolling 5-year percentile of manufacturing PMI - Adjust positions based on PMI levels: full position when >60%, empty position when <40%, half position when between 40%-60% - Formula: $ \text{PMI Timing Indicator} = \begin{cases} \text{Full Position} & \text{if PMI Percentile} > 60% \ \text{Empty Position} & \text{if PMI Percentile} < 40% \ \text{Half Position} & \text{if 40% \leq PMI Percentile \leq 60%} \end{cases} $ - Factor Evaluation: The factor has shown poor timing performance with an annualized return of only 0.41% since 2009 [6][13] Factor Backtesting Results 1. Manufacturing PMI Timing Strategy - Annualized Return: 0.41% [6][13] - Comparison with Benchmark: Underperformed the Wind All A Index annualized return of 8.49% [6][13] Style Rotation Models and Construction Methods 1. Model Name: Growth-Value Style Rotation Model - Model Construction Idea: The model suggests overweighting growth based on economic cycle analysis, valuation differences, and sentiment indicators [35][36] - Model Construction Process: - Analyze economic cycle indicators: profitability slope, interest rate cycle, credit cycle - Calculate valuation differences: PE and PB percentiles - Assess sentiment indicators: turnover and volatility differences - Formula: $ \text{Growth-Value Rotation Indicator} = \frac{\text{Profitability Slope Indicator} + \text{Interest Rate Cycle Indicator} + \text{Credit Cycle Indicator} + \text{PE Difference Indicator} + \text{PB Difference Indicator} + \text{Turnover Difference Indicator} + \text{Volatility Difference Indicator}}{7} $ - Model Evaluation: The model suggests overweighting growth based on current indicators [35][36] 2. Model Name: Small-Cap Large-Cap Style Rotation Model - Model Construction Idea: The model suggests balanced allocation based on economic cycle analysis, valuation differences, and sentiment indicators [35][41] - Model Construction Process: - Analyze economic cycle indicators: profitability slope, interest rate cycle, credit cycle - Calculate valuation differences: PE and PB percentiles - Assess sentiment indicators: turnover and volatility differences - Formula: $ \text{Small-Cap Large-Cap Rotation Indicator} = \frac{\text{Profitability Slope Indicator} + \text{Interest Rate Cycle Indicator} + \text{Credit Cycle Indicator} + \text{PE Difference Indicator} + \text{PB Difference Indicator} + \text{Turnover Difference Indicator} + \text{Volatility Difference Indicator}}{7} $ - Model Evaluation: The model suggests balanced allocation based on current indicators [35][41] 3. Composite Model: Four-Dimensional Style Rotation Model - Model Construction Idea: The model combines growth-value and small-cap large-cap rotation models for allocation [35][44] - Model Construction Process: - Combine the signals from growth-value and small-cap large-cap rotation models - Adjust positions based on combined signals - Formula: $ \text{Four-Dimensional Rotation Indicator} = \frac{\text{Growth-Value Rotation Indicator} + \text{Small-Cap Large-Cap Rotation Indicator}}{2} $ - Model Evaluation: The model suggests specific allocation proportions based on current indicators [35][44] Style Rotation Model Backtesting Results 1. Growth-Value Style Rotation Model - Annualized Return: 11.65% [35][37] - Comparison with Benchmark: Outperformed the benchmark annualized return of 6.91% [35][37] 2. Small-Cap Large-Cap Style Rotation Model - Annualized Return: 12.32% [35][42] - Comparison with Benchmark: Outperformed the benchmark annualized return of 7.11% [35][42] 3. Composite Model: Four-Dimensional Style Rotation Model - Annualized Return: 13.22% [35][44] - Comparison with Benchmark: Outperformed the benchmark annualized return of 7.50% [35][44]