The latest issue of Journal of Portfolio Management has a fascinating paper by Ziemba called "The Symmetric Downside-Risk Sharpe Ratio." The technical result is interesting but also interesting is that he shows the performance (in both numbers and graphs) of Soros, Buffet, Robertson, and Keynes(!).You can download the paper for free here: Ziemba

I didn't read much of the paper. But something bothers me about using a collection of outlier performances to compare performance metrics. Unless he thinks the performances are representative of the expectation of outlier managers rather than being outliers among similar managers. If that even makes sense.Which do you think is greater, the variability of performances, or the variability of managers generating the performances? Their could be few enough outlier managers, that in a given period none of them would achieve outlier performances, something like that...