However, as you know, since Dec '99 the overshoot has been substantial, with P/B ranging from 1.0 to 2.0.

Indeed. And 2007 especially.

My point is merely that if you look at the historical average of forward returns from any given valuation level, and thinking of that average as a forecast for the future, you are implicitly assuming that there WILL be an overshoot again, at least to some degree. I think a more parsimonious model is to assume a typical valuation level in future.

It took me some years to notice this flaw in what I had been doing...it didn't jump out until I had models of the two different styles which were giving very different expectations for the same forward time frame and built from the same amount of history and same starting valuation level. This led me to ponder which was likely better expectation, which I ultimately decided was "expect the typical". It's purely a bonus that it's dead easy. I have a spreadsheet with thousands of rows and now I only use four of them : )

Jim