Overnight news sentiment bias
factor.formula
Overnight News Sentiment Bias Factor:
in:
- :
Indicates the trading day corresponding to the calculated factor value. Specifically, the factor uses all news sentiment data in the time window [t-1 day 15:00, t day 9:25] to calculate the overnight sentiment bias factor on day t. The time window selection logic is: from the close of the previous day (15:00) to the opening of the next day (9:25), covering the news information during the non-trading period of the market, and the total number of news is N.
- :
They represent the probability that the sentiment analysis model determines that stock i has positive, negative, and neutral sentiment in the jth news in the overnight time window. These probability values reflect the intensity of the sentiment expressed by the news text, and the value range is [0,1].
- :
Represents the relevance score between stock i and the jth news. This score measures the relevance of the news content to a specific stock and is used to filter out news with a low relevance to the stock. The higher the relevance score, the greater the impact of the news on the stock. The value range is [0,1].
factor.explanation
This factor quantifies the sentiment of stocks affected by news during non-trading hours. By analyzing overnight news data and combining the positive, negative, and neutral sentiment of news and its correlation with stocks, it comprehensively evaluates the potential impact of market sentiment on stocks. Since the news information during this period has not been fully reflected in intraday trading, its sentiment may contain predictive power for future stock price changes. This factor assumes that overnight news sentiment that is not fully digested by the market has a positive impact on subsequent stock returns, and the correlation may be higher than that of intraday news sentiment.