S&P projections — statistics abuse
My brother put up a post on the other blog that involved a lot of handwaving around statistics. "That's absurd," I thought, "but I can top it." So I present you here with the results of my S&P500 extrapolations.

Based only on trailing price and earnings data, both adjusted for inflation, I've put together five and ten year projections. Both come in, as of late last year, at about 700 in January, 2000 dollars; i.e. by late 2008 and late 2013, the S&P is projected to have dropped by about 30% in real terms from where it is now. If you take to heart that this has significant uncertainty in it, it should be taken as broadly plausible; do not expect a reversion to '90's performance over the next decade.

Taking this outside the domain of its applicability, though, we find something much more interesting. I went back to see what would have been projected five and ten years from each month back into the twenties, and the ten-year projection from January of 1996 and the five-year from January 2001 — i.e. the two projections for January 2006 — are almost identical, at 1085 and 1089 in 2000 dollars. Both of them give a value in the next month that is some 7 to 8% lower; the projections rally gently through the summer of '05, roll over a top in the fall, and fall like a rock in 2006. The ten-year projection stabilizes in the spring of 2007, and it and the five-year are both about 20% lower than January 2006; at this point they part ways, as the ten-year stabilizes and recovers that 20%, while the five year drops another 20% by the end of the summer.

As I say, though, point by point movement-following isn't what this was designed for. That said, if we gently roll across a peak in the fall of 2005, I may be superstitious enough to be unusually well-hedged come the new year.
- posted by dWj @ 6:56 PM