This MSN Money article Why the market is melting down caught my attention because author Jon Markman makes a common statistical mistake in the justification of the theory presented. In the final paragraph he states
The average decline of the 14 bear markets since 1949 was 26.5%, ranging from 50.5% in 2002 to 8.3% in 1998. Right now, we're off about 5% from the early-May peak, so even if you pick the midpoint between the best at -8% and the average of -26%, it could be a pretty ugly result.
The problem with this analysis is that it’s data mining. He takes the average decline of bear markets. But the only way to get the average decline of a bear market is to (in hindsight) define somewhat arbitrarily a bear market as already being down a certain amount, ignoring all the times that the market stopped declining before that point and then turned around.
The appropriate question to ask statistically is what happens to markets after they’ve gone down 5% in a particular period of time, say one month?
Do they then tend to go down to his average bear market decline? Or do they tend more frequently to turn around and regain that ground?
I’m going to do this math. I haven’t yet, but I will and I’ll post more on it then. My suspicion is that roughly 50% of the time they turn around at that point or within 1% more of decline and that the market very rarely continues downward to the average bear market decline. I also suspect that the shorter the timeframe in which the market makes its decline, the more likely it is to bounce back quickly.