量化研究系列报告之二十一：ETF资金流透视：被动化浪潮下行业与个股的演进

Quantitative Models and Factor Construction Quantitative Factors and Construction Process 1. Factor Name: w_passvie_zs (Passive Holding Proportion Time-Series Change) - Construction Idea: Measures the time-series change in passive holding proportion using z-score standardization to eliminate differences in industry size and volatility[34][41][46] - Construction Process: 1. Calculate the passive holding proportion as the ratio of passive market value to daily free-float market value 2. Standardize the time-series change of the passive holding proportion using z-score[34][41][46] - Evaluation: Demonstrates strong positive predictive ability in 2024, with high excess returns in the long portfolio, especially in stable and incremental industries[40][44][143] 2. Factor Name: mv_passive_chg (Passive Holding Market Value Change Rate) - Construction Idea: Captures the rate of change in passive holding market value over a specific period, reflecting both ETF inflows/outflows and stock price changes[34][41][46] - Construction Process: 1. Calculate the change in passive holding market value over a given period 2. Divide the change by the initial passive holding market value to obtain the rate of change[34][41][46] - Evaluation: Transitioned from negative predictive ability during 2020-2023 to positive predictive ability in 2024, indicating a shift in market dynamics[119][144] 3. Factor Name: passive_slope (Passive Market Value Slope) - Construction Idea: Measures the slope of passive holding market value over time, combining price reversal effects with ETF fund flow dynamics[127][128] - Construction Process: 1. Calculate daily passive holding market value for each stock 2. Compute the slope of passive holding market value over 20, 40, and 60 trading days[127][128] - Evaluation: Exhibits stronger long-side stock selection ability compared to traditional reversal factors, with lower drawdowns and higher stability in long portfolios[128][130] 4. Factor Name: passive_beta (Passive Fund Flow Sensitivity) - Construction Idea: Measures the sensitivity of stock price changes to passive fund flow changes, reflecting the degree of synchronization between price and fund flow[131][132] - Construction Process: 1. Calculate daily stock price changes and passive holding market value changes 2. Compute the beta coefficient of stock price changes relative to passive fund flow changes over 20, 40, and 60 trading days[131][132] - Evaluation: Stocks with lower sensitivity (lower beta) exhibit better future performance, making this a long-biased factor[133][134][137] Factor Backtesting Results 1. w_passvie_zs - Rank IC: 2.23% - IC Win Rate: 62.7% - Long Portfolio Annualized Excess Return: 8.8% - Maximum Drawdown: -16.2%[46][121][122] 2. mv_passive_chg - Rank IC: 4.2% (2024) - IC Win Rate: 59.3% - Long Portfolio Annualized Excess Return: 2.2% - Maximum Drawdown: -11.9%[119][121][122] 3. passive_slope - Rank IC: -2.01% - IC Win Rate: 59.3% - Long Portfolio Annualized Excess Return: 2.9% - Maximum Drawdown: -13.1%[128][134][136] 4. passive_beta - Rank IC: -2.22% - IC Win Rate: 72.9% - Long Portfolio Annualized Excess Return: 4.7% - Maximum Drawdown: -8.2%[134][137][140]