>This should be fairly straight-forward to quantify. The number of
>stocks moving up vs. down, the number of stocks moving up more than the
>market average. The average size of stocks in both those categories,
>that kind of thing.
There aren't enough instances to do it rigorously. I'm talking about significant stock market tops, like the late 1920s and late 1960s/early 1970s, the kind with political and social consequences/associations.
Speaking of statistical confirmation, I just read a paper showing that for most stock markets outside the U.S., it's impossible to prove the long-term return is significantly different from 0. For the U.S. it most certainly is, though.
Doug