Originally Posted by LA Guy
i just wanted to point out that translating any concept from one culture to another is far from trivial.
I don't think that Fuuma would say that all quantitative modeling is stupid, but it's certainly worth nothing that such models generally require an expert user to properly apply them and properly intepret the results and uncertainties. It's also worth nothing that the more trivial the problem, the better models generally perform, which means that for very difficult, complex problems, a model is often as good as a shot in the dark. Why do you think companies adjust their growth and revenues projections constantly? Why do you think that companies fail in spectacular fashion, on a regular basis? The old saying "There are liars, damn liars, and then there are statisticians (or more accurately, data scientists)", is not a bad one to remember.
Also, greed can get in the way of reason. For example, it should have been foreseen that distributed risk does not mean that all risk therefore becomes infinitesimally small, but when people are making money hand over fist, all the blood streaming to the phallus makes for people to think.
I am indeed not dismissing quantitative modelling, merely pointing out that:
-Qualitative assumptions are often turned into quantitative data, which means you end up with "fake quantitative" data.
-Errors are compounded so you can end up waaaaaayyyy off
-Expertise is needed + if youre modelling for the future price of oil (I know a guy who does/did that, required quite the brainiac but was still way off) or something super complicated with a lot of geopolitical assumptions well....you're doing fake quantitative analysis.
-Decisions in companies are not taken through some sort of quant methodology, this is merely one of the myriads of help/hindrances to decision. Do not forget that the principle of most modern orgs, be they private or public, is to separate expertise from decision