>The premise behind SJIM is theoretically sound and can be boiled down to one observation: "the average stock picker performs horribly ."
No it is not at all theoretically sound. The biggest issue with stock picking is low volatility. This does not solve that. The other issue is that stock pickers don't do better or worse than a random selection of stocks after adjusting for factor tilt. That means stock pickers can't do better than average, but they also can't do worse than average - so the whole premise makes no sense if you actually buy into the idea that stock picking doesn't work.
Another theoretical issue is that UNDER performance doesn't mean NEGATIVE performance. If the market average is returning 10% one year, and a stock-picker returns 6% (maybe even 8% - but with fees eating 2%), then doing the OPPOSITE of the stock picker would be a bad idea. You don't want to short stocks that (on average) are having positive returns ...
>That means stock pickers can't do better than average, but they also can't do worse than average.
Not True. In fact virtually all stock pickers will do better or worse then the average. What you mean to say is the "average" stockpicker can't do better or worse then the average but that is of course a tautology.
What I actually mean is that 99.999% stock pickers can't intentionally do better or worse than average. And the tiny fraction who can are fund directors who are flooded with so much capital that they generate sub-1% alphas (which they eat up with their own fees).
Wouldn't it be possible to do worse? I think there should be enough opportunities to pick up true losers or almost dying companies.
I'm not saying you could extract profit for those situation as there are too many other people trying to do it. But losing money should certainly be possible.
1. You could try to do worse than average, and you'd succeed, but only because of the skewness of individual stock performance. You'd be no better than random chance at picking losers, it'd just be that most stocks are losers.
2. Factor-correction means the strategy of "choosing dying companies" won't work (e.g. in the Fama and French five factor model a dying company would an outlier in terms of high-minus-low, robust-minus-weak, and probably conservative-minus-aggressive).
No it is not at all theoretically sound. The biggest issue with stock picking is low volatility. This does not solve that. The other issue is that stock pickers don't do better or worse than a random selection of stocks after adjusting for factor tilt. That means stock pickers can't do better than average, but they also can't do worse than average - so the whole premise makes no sense if you actually buy into the idea that stock picking doesn't work.