05 August 2019

Probably? Not

It's no secret that I've some problem using pure quant in the arena of human events. The cause and effect of human events are determined, almost entirely, by the rules surrounding said events. And said rules can and will be changed even in-flight if those who seek to benefit from said events see that the desired outcome is not forthcoming. I'm talking to you, Blythe Masters. Time series analysis to predict booms and busts in the economy principally among them. Given The Manchurian President's behavior, would anyone be surprised if he'd managed to threaten BLS and Census Supergrades (aka, SES)? Of course he's tried. Could the really, really rosy numbers coming out of BLS be fudged? If the Secretary of an agency can clear out the Supergrades with a snap of the fingers, what do you think?

Well, here are a couple of reports that might be of interest to those who share my chary attitude.

Should Probabilistic Inference be Thrown Under the Bus?
The real problem is that entire practice of Bayesian inference, and its Information Theory relative, is surprisingly fundamentally flawed in non-linear domains.

The second is older, and closer to the Great Recession "Kill the Quants?": Why risk analysis fails.
The financial crisis featured both models that were wrong and models that were right but were ignored by managers. Modeling errors vied with deliberate misstatements. Federal agencies deferred to the expertise and presumed non-malevolence of private firms.

I really like "presumed non-malevolence of private firms". As if!! Blythe Masters, Blythe Masters, Blythe Masters

No comments: