But it's interesting that decades after the original value investing texts were first published, only a handful of investors have managed to successfully beat the markets--and those who do often employ similar strategies that implicitly weed out the majority of stocks as decently priced. To me, that implies that just a few people are endowed with the capacity to do something different (They have a different emotional mindset?) while others can't copy. This is akin to having private information. Their presence will make the markets more efficient, but it can take some time and they'll make a good deal of money as they do so.
Portfolio Theory, however, has never made much sense to me. There is of course a resemblance to Efficient Markets, but the assumptions are crazy. For one, you have no accounting for Talebian Black Swan risk, which has taken out many people including LTCM. There's no economic reason that past performance predicts future extreme risk--an event which is by definition low probability. So-called post-modern Portfolio Theory is a little nicer as it correctly terms risk as the downward portion of volatility, not uncertainty itself. But another approach is to throw out normalities altogether and focus on fractals, as Mandelbrot does. I've often thought stock charts appear self-similar at different time scales, and Mandelbrot takes this approach to produce a pattern which echoes Portfolio Theory's randomness for much of the time but with a healthy appreciation for crazy events. It's promising stuff, and not only because it involves fractals.