Intraday information asymmetry intensity factor
factor.formula
1. Calculate daily return:
a. Overnight return of individual stocks:
b. Index overnight return:
c. The daily afternoon stock returns:
d. Daily afternoon index return:
2. Calculate the residual term:
a. Perform a linear regression on the overnight returns of individual stocks and the index to eliminate the impact of overall market volatility:
Get the overnight return residual:
b. Perform a linear regression on the afternoon returns of individual stocks and the index to eliminate the impact of overall market volatility:
Get the afternoon profit residual:
3. Calculate the difference between the daily overnight and afternoon residuals:
4. Construct statistics stat:
a. Calculate the mean of δ:
b. Calculate the standard deviation of δ:
c. Using the above means and standard deviations, calculate the T statistic to measure the significance of the difference between the overnight and afternoon return residuals:
5. Eliminate the impact of momentum factor:
a. Perform a cross-sectional regression of the statistic stat on the momentum factor (the past 20-day return) to eliminate the momentum effect:
b. The residual term ε obtained from the regression is the intraday information asymmetry intensity factor.
in:
- :
The overnight return rate of an individual stock on the i-th day usually refers to the return rate of the opening price of the day relative to the closing price of the previous day. A more precise definition can be adjusted based on actual trading data.
- :
The overnight return of the index on the i-th day corresponds to the overnight return of the individual stock, and the definition should remain consistent.
- :
The return rate of an individual stock in the afternoon session on the i-th day is usually defined as the price return rate from the opening to the closing of the afternoon session. The specific time period definition needs to be precisely adjusted according to the actual trading time of the exchange.
- :
The return rate of the index in the afternoon on the i-th day corresponds to the return rate of the individual stock in the afternoon, and the definition should remain consistent.
- :
The intercept term in linear regression represents the expected value of individual stock returns when the market index return is zero.
- :
The regression coefficient in linear regression indicates the expected change in individual stock returns when the index return changes by one unit.
- :
The residual term of the overnight return regression model of individual stocks on the i-th day represents the part of the overnight return of individual stocks that cannot be explained by the model. This residual term can be understood as the overnight return information of individual stocks after eliminating the market impact.
- :
The residual term of the regression model of the afternoon return of individual stocks on the i-th day represents the part of the afternoon return of individual stocks that cannot be explained by the model, which can be understood as the afternoon return information of individual stocks after eliminating the market impact.
- :
The difference between the overnight return residual and the afternoon return residual on day i is used to capture the difference in intraday return patterns, which may reflect the degree of intraday information asymmetry.
- :
The mean of δ represents the average level of the difference between overnight and afternoon return residuals over a period of time (e.g., N days), representing the overall intraday return pattern.
- :
The standard deviation of δ measures the volatility of the difference between the overnight and afternoon return residuals over a period of time, reflecting the stability of the intraday return pattern.
- :
The sample size used when calculating the mean and standard deviation, usually refers to the size of the time window. The larger the sample size, the more stable the statistical results.
- :
The T statistic for stock j is used to assess the significance of the difference between the overnight and afternoon return residuals. The larger the absolute value, the more significant the difference, which may indicate that the stock is more strongly affected by information asymmetry.
- :
The return rate of stock j in the past 20 trading days is used to control the impact of momentum effect. Here, 20 days is a commonly used window period and can be adjusted according to actual conditions.
- :
The regression residual of stock j represents the unique return signal caused by intraday information asymmetry after removing the momentum effect. This value is the final intraday information asymmetry intensity factor value.
factor.explanation
This factor is based on the assumption that informed traders are more active in the morning, and constructs a quantitative indicator to characterize the degree of information asymmetry within the stock market. The improved APM factor uses overnight returns instead of morning returns to better capture the impact of pre-market information disclosure on prices. Market volatility is eliminated through linear regression, and the T statistic is combined to measure the significance of the difference between the residuals of overnight and afternoon returns. The momentum effect is further eliminated through cross-sectional regression to obtain the final intraday information asymmetry intensity factor, which can be used in quantitative stock selection strategies to help identify stocks that may have information advantages. This factor can be used in combination with other fundamental factors, technical factors, etc. to construct a multi-factor model.