Followed Clif High's webbot project for about year (subscription to monthly reports that were ~20 pp. long -- still have them on hard drive) and it was interesting ... for a while ... then he seemed to go off into a ditch and started promoting bitcoin (which he obviously was mining in a big way) with a much higher price subscription service devoted to "predicting" what bitcoin would do. In other words an investment letter for those into bitcoin. I bailed at that point since I judge bitcoin and all cryptocoins to be a very unsafe, risky investment, a form of Ponzi scheme subject to hacking and government intrusion, not capable of meeting any of its promises that the faithful believe in. (Time will tell.)
High's methology of mining the web for linguistic data then assigning values and association to linguistic terms probably did rake in some very interesting stuff that people were talking about or thinking about. His precise methodology was never disclosed to the public. But on top of the linguistic data, he added his own interpretations (predictions) each month, which seldom turned out to be correct.
He did turn me on to Antarctica being a very spoopy place. And the reality of a Secret Space Program.
I do believe that predictive analytics is a THING and that a combination of very massive databases with massive computing power, and now possibly with the use of quantum computing -- which can rapidly do the probability calculations at a speed that no conventional computer can match no matter how large -- is involved.
I would assume that governments -- particularly OUR government -- have the very best capabilities for predictive analytics. I also assume that some other organizations and governments are doing it.