Market-neutral turnover rate residual
factor.formula
Cross-sectional regression model:
in:
- :
is the natural logarithm of the daily average turnover rate of stock i in the past month in period t. The turnover rate is calculated as the daily trading volume divided by the total outstanding shares, and then the daily average within a month is taken. The logarithm is taken to reduce the skewness of the turnover rate distribution, making it closer to the normal distribution, which is conducive to regression analysis.
- :
is the natural logarithm of the circulating market value of stock i in period t. The circulating market value usually refers to the market value of stocks that can be freely traded, and is calculated by multiplying the stock price by the circulating shares. Similarly, taking the logarithm is to reduce the skewness of the market value distribution and make it closer to the normal distribution.
- :
It is the intercept term of the regression model in the tth period, representing the expected logarithm of the turnover rate when the market value is 0.
- :
is the slope coefficient of the regression model in period t, representing the marginal impact of stock market value on turnover rate, which is usually negative. This coefficient reflects the negative correlation between market value and turnover rate, that is, stocks with larger market value tend to have lower turnover rate.
- :
is the regression residual term of stock i in period t, which is the value of this factor. It represents the fluctuation of turnover rate unique to stock i after removing the influence of market value. The larger the residual, the higher the turnover rate of the stock relative to its market value, and vice versa.
factor.explanation
This factor is based on a reasonable assumption: the turnover rate of stocks is largely driven by market capitalization, that is, large-cap stocks tend to trade more quietly. Therefore, we use a cross-sectional regression model to separate the market capitalization-related part of the turnover rate, and the resulting residual \epsilon_{i,t} represents the specific turnover rate change independent of the market capitalization. We believe that stocks with larger (positive) residuals are more likely to be noticed by the market and thus obtain excess returns; stocks with smaller (negative) residuals may be traded coldly and have poor returns. This market capitalization-neutral treatment helps improve the performance of the turnover rate factor in long portfolios. Compared with directly using the turnover rate, the market capitalization-neutral treatment can more accurately capture the signals in the turnover rate related to stock fundamentals, thereby improving the effectiveness of the factor.