The real question is whether the hypothesis should be built on the
underperformance of the majority or on the outperformance of the minority
(if it is strongly clustered)
More specifically, the question is whether the overperformance of the minority could have been predicted based on information which was publicly available at the times they traded, since they were presumably not privy to inside information that wasn't also available to the rest of us. This question gets at the gist of market efficiency. Unfortunately, we don't understand the trading rules of these successful traders (and probably never will).
Alternatively, we might ask whether these traders have discovered ways of reducing transaction costs that the rest of us haven't caught onto, allowing them to profit in ways other informed investors can't.
Also, as I teach my students, market efficiency is an equilibrium concept. If someone is not there to take advantage of new information first, then "efficient" prices will never obtain. Are these successful traders the elusive "information arbitrageurs" that make markets efficient? I suspect not since, if I were one of them, I wouldn't want you to know!
Finally, and probably most to the point, how are we measuring "outperforming the market"? This is a crucial issue in empirical studies of market efficiency. Since the profits of the traders we're discussing were reported in a case study, it's not clear that any rigor was used in measuring abnormal profit. If these investors specialized in portfolios whose risk characteristics were greater than those of the overall market, we expect them to earn a higher rate of return than that of a market portfolio. So, how carefully was this done. Even rigorous studies of market efficiency are sensitive to the problem of mis-specification of the assumed asset pricing model -- the model upon which abnormal profits are measured. In fact, given our current scientific technology, market efficiency cannot be empirically refuted since we cannot distinguish a genuine inefficiency from a mis-specified asset pricing model. Unfortunately, all empirical tests (except experiments) jointly test efficiency and the asset pricing model assumed.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Kevin D. Sachs, Ph.D.
Assistant Professor
University of Cincinnati
Department of Accounting/IS
302 Lindner Hall, P.O.Box 210211
Cincinnati, OH 45221-0211
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
