News co-occurrence attention overflow
factor.formula
The proportion of co-occurring news between stock i and stock j on trading day t:
The month-on-month change in the proportion of news items that appeared together between stock i and stock j on trading day t:
News co-occurrence attention spillover intensity of stock i on trading day t:
in:
- :
is the number of times (or number of news articles) that stock i and stock j appear together in news reports within trading day t
- :
is the total number of times (or number of news articles) stock i appears in news reports alone or together with other stocks within t trading days.
- :
is the ratio of the number of co-occurring news about stock i and stock j to the number of all news about stock i within trading day t, which measures the co-occurrence strength between stock i and stock j
- :
is the month-on-month change in the proportion of news items co-occurring between stock i and stock j on trading day t, measuring the change in the co-occurrence intensity between stock i and stock j
- :
It is the sum of the month-on-month changes in the proportion of news co-occurrences between stock i and all other stocks on trading day t, which measures the intensity of news co-occurrence attention spillover received by stock i on trading day t. Note that the summation range of j here does not include i itself to avoid repeated counting.
factor.explanation
This factor measures the market attention spillover effect by calculating the change in the intensity of news co-occurrence between stocks. When the co-occurrence frequency of stock i with other stocks increases, it means that the investor attention of stock i may be distracted, triggering additional market attention to it, which may cause irrational fluctuations in investor sentiment. A higher factor value may indicate that the stock may be over-focused and may perform poorly in the future. Conversely, a lower factor value may indicate that the stock's attention is declining and its future performance may be relatively stable.
In practical applications, the following points can be considered:
- Time window: According to actual needs, the time window for calculating the co-occurrence frequency can be adjusted, such as weekly, monthly, etc.
- News source: Different news sources can be selected according to actual conditions, such as news websites, social media, etc., and the weight of the news source can be considered.
- Industry/sector: The co-occurrence effect within the industry or sector can be considered. For example, the co-occurrence of stocks in the same industry may be more indicative.
- Combined with other factors: This factor can be combined with other factors to build a more robust portfolio model.