I used both Python and Ruby extensively for a while, love both, but Ruby was just nicer to code in. The problem for me was the much smaller ecosystem, especially for anything numerical/data related. Numpy was, pre Pandas days, the killer library for the type of work I do.
The headline itself demonstrates ignorance as to what matters to an investor putting money on the table.
1. Sell-side analyst forecasts do not drive the investment decisions of the more sophisticated investors. The sell-side is just a big marketing machine and the value-add of the analysis is very low at an individual stock level.
2. The model ‘beats’ the consensus forecast on a limited sample of names. By definition, the consensus forecast is an averaging out which leads to dilution of any one analyst’s alpha - hence it is not an appropriate benchmark.
It is a naive approach and study, but typical of academics who unfortunately have little exposure to real-world investing/trading strategies.
The use of ‘alternative data’ is not new and is definitely leading to alpha generation for some firms, but as mentioned by others such data-driven strategies will usually have limited shelf-life.
Thank you, Paul Graham, on behalf of obsessive collectors everywhere! This also supports the ‘no free-lunch’ adage. There is merit in exploring uncharted territory, purely for the sake of it, but it is hard work with no guarantees of finding treasure except satisfaction of doing what you are interested in.
