Lazy Prices [1] is a paper I was looking at for a while (I work on a Quant Research team) that does this in a simple way and the authors had good results. It takes the changes in language in some key sections of financial fillings as signals.

[1] https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1658471

Thanks for pointing to the paper, haven't read it yet.

But it sounds pretty interesting, especially if you correlate it with things like insider's tradings (both buy and sell) and of course whatever external events there may be.

Talking of quant, would you mind sharing what is your typical stack, especially non-proprietary, non-confidential, stuff such as languages and libraries.

I don't do anything low latency so its mostly the common scientific python stack -- numpy, pandas, sklearn, etc. Mostly linux, lots of Spark, some R and C#. We're really flexible on what we can use to solve our problems but the team has pretty much settled on python as both developers and analysts can work together more easily.

If I recollect right, Prof. Damodharan has something like a CHINA score - counting the number of mentions of China in S1 or quarterly reports as a proxy for bullshit.

This is one of the types of things quants at hedge funds do.