量化选股因子跟踪月报：上月预期、成长和质量因子表现较优-20250901

• The report tracks the performance of 48 representative factors across 12 major styles, including scale, Beta, volatility, value, liquidity, momentum & reversal, technical, profitability, growth, quality, dividend, and consensus expectations. These factors were selected based on their relative performance over a 10-year backtest period[19][20] - Factor data preprocessing includes outlier removal, industry and market capitalization neutralization, and z-score standardization. For industry neutralization, OLS regression with industry dummy variables is used, while market capitalization neutralization involves regression with logarithmic market capitalization. The residuals from these regressions represent the neutralized factor values[199][202][203] - IC analysis measures the correlation between factor exposure and future stock returns using Spearman rank correlation coefficients. Positive IC values indicate logical and expected factor performance. Factors are also tested through layered backtesting and regression analysis to evaluate their effectiveness[204][205][206] - Layered backtesting involves sorting stocks by factor scores, dividing them into layers, and observing the cumulative returns of each layer. This method evaluates the linear and non-linear relationships between factors and stock returns[206][207] - Regression analysis controls for industry and market capitalization effects, using a linear model to assess the relationship between factor exposure and stock returns. The regression coefficients reflect the factor's predictive power, with significant t-values indicating robust factor performance[207][208] - The report highlights the monthly performance of factors across different stock pools (Wind All A, CSI 300, CSI 500, CSI 1000). Factors such as growth, quality, and consensus expectations showed strong performance, while volatility and liquidity factors experienced significant drawdowns[2][3][21] - Specific factors like turnover rate standard deviation (1-month), reversal (1-month), and turnover rate-price correlation (1-month) performed well among volume-price factors. Financial factors such as quarterly revenue growth and return on invested capital (ROIC) also showed notable performance[4] - In the Wind All A stock pool, the expectation factor achieved an IC of 5.19%, a long-short return of 1.37%, and a long-only excess return of 0.32%. The reversal factor had an IC of 4.83%, a long-short return of -0.65%, and a long-only excess return of 0.37%[2][21] - In the CSI 300 stock pool, the expectation factor achieved an IC of 25.94%, a long-short return of 11.07%, and a long-only excess return of 3.44%. The quality factor had an IC of 18.24%, a long-short return of 4.64%, and a long-only excess return of 1.73%[2][21] - In the CSI 500 stock pool, the growth factor achieved an IC of 0.70%, a long-short return of 2.90%, and a long-only excess return of 1.24%. The expectation factor had an IC of 0.50%, a long-short return of 2.01%, and a long-only excess return of 0.93%[2][21] - In the CSI 1000 stock pool, the technical factor achieved an IC of 2.02%, a long-short return of -1.27%, and a long-only excess return of 0.02%. The growth factor had an IC of 0.88%, a long-short return of 1.68%, and a long-only excess return of 1.13%[3][21] - Factors such as Beta, small-cap, and volatility showed negative performance overall, with significant drawdowns in specific stock pools. Dividend and value factors also underperformed across all stock pools during the month[3][46][97][174] - The expectation factor demonstrated strong performance in large-cap stock pools, particularly in the CSI 300 pool, with positive excess returns and long-short returns. Growth factors showed consistent positive performance across all stock pools, with more pronounced results in large-cap pools[3][143][186]