Yahoo finance conveniently provides historic ratings from a number of analysts. Unfortunately, each analyst seems to use a different scale: buy/hold/sell, perform/outperform/neutral, overweight/equal weight/underweight, etc.
I've already written a function to scrape this data into R, and I would like to expand it by automatically coding the ratings. I was thinking of using a -1,0,1 system, which would basically encode bad/neutral/good, and trying to shoehorn each analyst's system onto this scale.
I'm sure other people have solutions to this problem, and I'm interested to see what they are before I start working.