红利风格投资价值跟踪（2025W22）：美联储9月降息概率走高，红利ETF本周转为净流入

• The macro model uses indicators such as "10-year US Treasury yield," "domestic M2 year-on-year growth," and "domestic M1-M2 scissors difference" to predict excess returns of the CSI Dividend Index relative to Wind All A Index. The model has achieved an annualized excess return of 8.05% since 2010, but the excess return for this year is -6.69%[8][10][46] - The valuation model employs a weighted factor adjustment for the CSI Dividend Index's PETTM valuation to align with its dividend yield-weighted characteristics. The absolute PETTM valuation regression formula is $y = -0.2807x + 0.2635$, where $x$ represents the three-year absolute PETTM percentile, and $y$ represents the future one-year absolute return. The relative PETTM valuation regression formula is $y = -0.1223x + 0.0983$, where $x$ represents the three-year relative PETTM percentile, and $y$ represents the future one-year excess return[18][21][30] - The price-volume model calculates the weight of CSI Dividend Index constituents above the 120-day moving average and their correlation with future one-year absolute returns. The regression formula is $y = -0.2346x + 0.2116$, where $x$ represents the weight above the 120-day moving average, and $y$ represents the future one-year absolute return. For absolute trading volume, the regression formula is $y = -0.3823x + 0.3435$, where $x$ represents the three-year trading volume percentile, and $y$ represents the future one-year absolute return. For relative trading volume, the regression formula is $y = -0.0163x + 0.0092$, where $x$ represents the three-year relative trading volume percentile, and $y$ represents the future one-month excess return[24][31][33] - The fund exposure model standardizes quarterly dividend yield factors and calculates the dividend style exposure of equity-focused public funds based on their top 10 holdings. The model shows that the dividend style exposure for Q1 2025 is 0.37, a decrease from 0.45 in Q4 2024[37][38][46] - The "Dividend 50 Optimal Portfolio" combines high dividend yield with a linear multi-factor model to enhance capital gains. It applies Barra style factor constraints to control portfolio style exposure and incorporates a three-dimensional dividend excess timing model for further optimization. The portfolio's absolute return over the past year is 5.69%, with an excess return of 3.72%[46][47][48] Model Testing Results - Macro model: Annualized excess return since 2010 is 8.05%; excess return for this year is -6.69%[8][10][46] - Valuation model: Absolute PETTM regression predicts future one-year absolute return of -1.42%; relative PETTM regression predicts future one-year excess return of 0.00%[18][21][30] - Price-volume model: Weight above 120-day moving average regression predicts future one-year absolute return of 7.35%; absolute trading volume regression predicts future one-year absolute return of 13.78%; relative trading volume regression predicts future one-month excess return of 0.80%[24][31][33] - Fund exposure model: Dividend style exposure for Q1 2025 is 0.37, down from 0.45 in Q4 2024[37][38][46] - Dividend 50 Optimal Portfolio: Absolute return over the past year is 5.69%; excess return is 3.72%[46][47][48]