Two-way price difference autocorrelation normalization factor
factor.formula
CDPDP:
in:
- :
The first-order difference of the price at the tth time point is calculated as $\Delta P_t = P_t - P_{t-1}$, where $P_t$ is the price at time t.
- :
Positive price difference autocorrelation means taking the average of the 20-day correlation coefficient of the series consisting of $\Delta P_t$ and $\Delta P_{t+1}$ when the price difference is greater than zero (i.e. $\Delta P_t > 0$). This value measures the persistence of price increases.
- :
Negative price difference autocorrelation means taking the average of the 20-day correlation coefficient of the series consisting of $\Delta P_t$ and $\Delta P_{t+1}$ when the price difference is less than zero (i.e. $\Delta P_t < 0$). This value measures the persistence of the price decline.
- :
Represents an averaging operation, which is used to calculate the mean of positive and negative autocorrelations.
- :
Represents the standard deviation operation, which is used to calculate the standard deviation of positive and negative autocorrelations for standardization.
factor.explanation
The logic of this factor is based on the mean reversion characteristics of prices, and uses the method of double sequence difference to enhance its ability to capture reversal signals. When stock prices continuously change in the same direction, the value of this factor will be higher, and vice versa. Standardization ensures that the factors are comparable between different stocks. Therefore, stocks with low factor values indicate that the direction of price changes may reverse, which is usually considered a potential buying opportunity, and vice versa, it may be a selling opportunity. This factor has similar logic to the single sequence difference autocorrelation factor, but by calculating the positive and negative price difference autocorrelations separately, it enhances the ability to capture the reversal of price changes.