Over what time frame? Medallion over 12 years is supposed to be around 1.7 but I haven't actually run the numbers off of his report myself. Any guy with big assets breaking above a Sharpe of 1 for a decade is going to attract a horde of money, anyone doing 1.5+ over a decade is pretty much a legend. Could they optimize for a better Sharpe, I'm sure, but as you said before it is but a piece of the puzzle and you might be hurting overall performance by doing so. Sharpe is fairly important to these guys though as it is a rule-of-thumb / at a glance type of figure that institutions peruse. I suspect if your Sharpe is a .13 you might have a bit of difficulty attracting money even if you have some crazy annual returns.
Medallion does not take outside money. I think just like HFT firms they are capacity constrained. It is not like they do not want to have a bit lower sharpe but much more size. It is not feasible. Sharpe is capturing first 2 moments. More moments tell you more. Keep in mind if you know all the moments of a distribution, you can back out the entire distribution. The trouble of calculating higher moments are that you need lots of sample points to have a decent representation of the tail events.
atticus, I understood sharpe for selling vol or selling skew is not a good measure since you blow up acct when the tail events hit you. Where do you think the sharpe will be if selling iron condor/butterfly or calender version of it? the risk is now limited. njrookie
Hard to say. The DD will be lower (contract for contract) but haircuts will be significantly reduced, allowing the manager to risk more with the fly, diagonal, etc.
Victor Niederhoffer is a good example. http://www.newyorker.com/reporting/2007/10/15/071015fa_fact_cassidy
The importance of a 2 month Sharpe, are you kidding? I'd rather be concerned if the Sharpe is that high as it will come down rapidly to more normal levels.