Very interesting Paper.
Any way we can group stocks based on analysts that cover the same stocks (now or via a feature
request?).
The explanation on why this alpha exists is very interesting. Analysts specialize on stocks
that have commonality (industry, along the supply chain etc.)
Basically when an analyst updates an estimate positively, there is a time lag to updating
of stocks the analyst covers as well.
Therefore time laged (thats what we want :-)) momentum spill over happens to the other stocks the analyst covers.
Whats interesting that the 4 Week Industry Momentum is a great proxy for the co-coverage alpha effect. (“For example, the coefficient on past one-month industry return decreases by 69% once past CF return is added to the regression.”).
Best Regards
Andreas
4 Likes
Fascinating article, thanks for sharing Andreas. I look forward to reviewing in more detail. The abstract reminds me of another interesting quantitative research paper published by Morgan Stanley. In short, their team uses NLP (Natural Language Processing) to review each newly published analyst report on individual stocks. It then scores each research report based on the sentiment in the language used, weighting more recently published reports higher. It then groups the stocks by industry and assigns it a composite industry sentiment score + a momentum score. Going long the top and short the bottom industries generates significant alpha.
Their simple long / short strategy generates a Sharpe of 1.74
This is an interesting approach but one that needs significant infrastructure to implement IMO.