I must admit that it is mostly art to me
There is certainly some art in the practice of science, particularly in the intuition of directions to investigate and in forming of hypotheses with a good chance of survival. There are other places where art enters, but one would rather find something more dependable and so one hunts for what might have underlain the intuition, but which might be, at least in part, measurable and testable.
In the current topic, which aspects do you see as requiring or [currently] relying on art?
what would you use as MSFT's forward growth rate for 1980?
That would depend on the theory under test. If the theory was basing forward growth rates on analyst estimates, then one would want historical analyst estimates. These should be a matter of record, but I don't know where one would dig them out. If one were using growth rates based on instanteous rate as per the financials, well then that is certainly computable now as then.
To be sure, there are some difficulties in retro testing in trying to put oneself fully into the mood and perception one had at the time, but I don't think it renders the process impossible. Perhaps it is likely to render the test more likely to produce good results since one might use one's knowledge of history to influence the guesses back then, but whatever choices one makes, if one documents them, then someone else can evaluate, challenge, and make their own tests. |
