The goal of this newsletter is simple: collect as many hedge fund letters to investors as I can get my hands on and find consensus topics of discussion.
This is the initial prototype issue for the newsletter. A total of 41 letters were used for this report. Forty-one letters is too small a number to extract a real consensus, but I hope to acquire a much larger number of letters for future issues.
Letters are first annotated and undergo natural language processing (NLP) to identity named entities and coreferences (among other tasks) and are then pumped into a text search engine to find consensus topics among all funds. This same text search engine is available to readers to use for possibly finding additional topics of interest. Part of the annotation includes identifying new buys/sells, and existing holdings, fund performance, cash allocation, and key opinions.