Back in the Stone Age, when I was in business school, I took an Econometrics course–on a mainframe! I learned two things: the most accurate predictions came from autoregression, meaning projecting forward the past behavior of a given variable (rather than having a whole bunch of them); and there’s no failsafe warning if you accidentally hit the delete button on the keyboard.
The Financial Times had an article the other day by John Plender arguing a variation of the same thing dominates the world stock markets in the present day–that today’s computer-driven trading, as well as what’s done in traditional fashion by humans, both act as if today’s winners will be tomorrow’s and that recent market direction (right now meaning down) will continue.
The result is to make a trend reversal much less likely than it has been in the past.
I think some version of this idea is actually the case. If so, the question of the current selling trend exhausting itself is less one of market level than one of time and of the pool of potential sellers drying up. The end of the summer?