风格轮动择时周报：后市或有震荡，但不改中期向上观点-20260301

Quantitative Models and Construction Methods 1. Model Name: Growth-Value Rotation Model - Model Construction Idea: The model captures the switching rhythm between growth and value styles by using dividend low volatility and the ChiNext Composite Index as proxies for value and growth styles, respectively [3][14] - Model Construction Process: - The model uses historical data to identify the switching points between growth and value styles - It evaluates the relative attractiveness of the two styles based on their historical performance and other market indicators - The model then allocates weights dynamically between the two styles to maximize returns [3][14] - Model Evaluation: The model demonstrates the ability to effectively capture the switching rhythm between growth and value styles, achieving significant excess returns in the out-of-sample period [3][14] 2. Model Name: Large-Cap vs. Small-Cap Rotation Model - Model Construction Idea: The model identifies the rotation between large-cap and small-cap styles using CSI 300 and CSI 2000 as proxies for large-cap and small-cap styles, respectively [3][14] - Model Construction Process: - The model evaluates the relative attractiveness of large-cap and small-cap styles based on their historical performance and other market indicators - It dynamically allocates weights between the two styles to optimize returns [3][14] - Model Evaluation: The model shows effective generalization in capturing the rotation between large-cap and small-cap styles, achieving robust excess returns in the out-of-sample period [3][14] 3. Model Name: Dynamic Time Warping (DTW) Timing Model - Model Construction Idea: The DTW timing model is based on similarity analysis, comparing current market index trends with historical trends to identify similar patterns and generate timing signals [12] - Model Construction Process: - Calculate the similarity between the current index trend and historical trends using the DTW algorithm - Select historical segments with high similarity as references - Compute the weighted average future returns and standard deviations of the selected historical segments, using the inverse of the distance as weights - Generate trading signals based on the average future returns and standard deviations [12] --- Model Backtesting Results 1. Growth-Value Rotation Model - Annualized Return: 19.41% - Annualized Volatility: 23.63% - Maximum Drawdown: 29.09% - Sharpe Ratio: 0.82 - Return-to-Drawdown Ratio: 0.67 - Out-of-Sample Return (since 2024): 86.53% - 2025 Return: 60.89% - 2026 Return: 5.45% [19] 2. Large-Cap vs. Small-Cap Rotation Model - Annualized Return: 22.05% - Annualized Volatility: 23.19% - Maximum Drawdown: 28.66% - Sharpe Ratio: 0.95 - Return-to-Drawdown Ratio: 0.77 - Out-of-Sample Return (since November 2025): 14.68% - 2025 Return: 51.47% - 2026 Return: 12.95% [24] 3. DTW Timing Model - Latest Signal for CSI 300: Optimistic (Average Return: 0.13%, Standard Deviation: 0.06%) - Latest Signal for CSI 2000: Neutral (Average Return: -0.61%, Standard Deviation: 0.08%) - Latest Signal for ChiNext Index: No Signal (Average Return: -0.69%, Standard Deviation: 0.13%) [13] --- Quantitative Factors and Construction Methods 1. Factor Name: Style Odds - Factor Construction Idea: The relative valuation levels of market styles are key determinants of their expected odds, with a negative correlation between valuation levels and odds [8] - Factor Construction Process: - Calculate the valuation differential percentile between styles - Estimate the odds for value vs. growth styles and large-cap vs. small-cap styles based on the valuation differential percentile - Current odds: Value vs. Growth = 1.23, Large-Cap vs. Small-Cap = 1.68 [8] 2. Factor Name: Style Win Rate - Factor Construction Idea: The win rate of a style is determined by various market indicators, such as bond yields, market trends, and momentum [10][11] - Factor Construction Process: - Evaluate the win rate of each style based on a set of market indicators - Current win rates: Growth = 56.25%, Value = 43.75%; Large-Cap = 14.29%, Small-Cap = 85.71% [10][11] 3. Factor Name: Style Scores - Factor Construction Idea: Style scores are calculated based on the combination of odds and win rates to determine the attractiveness of each style [11] - Factor Construction Process: - Use the formula: $ \text{Investment Weight Score} = \frac{\text{Win Rate} \times \text{Odds} - (1 - \text{Win Rate})}{\text{Odds}} $ - Current scores: Large-Cap = -0.37, Small-Cap = 0.62, Growth = 0.02, Value = -0.02 [11] --- Factor Backtesting Results 1. Style Odds - Value vs. Growth Odds: 1.23 - Large-Cap vs. Small-Cap Odds: 1.68 [8] 2. Style Win Rate - Growth Win Rate: 56.25% - Value Win Rate: 43.75% - Large-Cap Win Rate: 14.29% - Small-Cap Win Rate: 85.71% [10][11] 3. Style Scores - Large-Cap Score: -0.37 - Small-Cap Score: 0.62 - Growth Score: 0.02 - Value Score: -0.02 [11]