Quote:
Originally Posted by ToothSoother
Given that there are thousands of hedge funds, is there any evidence that the standout hedge funds outperform beyond statistical significance?
I have been long threatening, but failing to deliver, on a post detailing the record of Millennium Management (I do not work there). Here goes:
A Sharpe ratio (named after economist William Sharpe) is defined as
(return - risk free rate)/(volatility)
put another way, excess return/vol
While there are other ways to measure and discuss performance, this is a simple and straightforward one and is often used.
Sharpe ratios are most commonly discussed in terms of annual returns and annual vols.
If you have a Sharpe ratio of 1, you would expect to beat the risk free rate approx 5 out of 6 years (DUCY?)
If you have a Sharpe ratio of 2, you would expect to make money approx 43 of 44 years.
If you are just taking random risks (no edge), the odds that you would have a realized Sharpe ratio of 1 or higher in any one year period is approx 17% and less than 3% of over 2. Could happen though...there are a lot of hedge funds....
The odds of a random risk taker having a 2 Sharpe for 4 years is now 4SDs (2 * (sqrt(4)) or approx 1 in 15787
Millennium has has a 2.65 Sharpe for 16 years. Over 10SDs. Wikipedia stops at 7SDs (1 in 390billion)
There aren't that many hedge funds!
(and by the way, these are their returns AFTER fees. It would be their before fee performance you would want if were an academic debating whether markets are efficient or not)
But what makes Millennium track record even more interesting is how they do it.
Their model is to hire an army of different risk takers, screened by the firm to those most likely to have edge. They give these risk takers an initial amount of capital. If they do well, they are given more capital. If they lose money they have capital removed or are asked to leave entirely. They are asked NOT to coordinate with each other in order to prevent group think from making the teams correlated to one another.
It is basically a genetic algorithm for determining whether edge exists or not!
In the HSBC report there are many hedge funds with long track records at 1 or higher Sharpe ratios. Taking their Sharpe times the square root of track record length in years, you will find an awful lot of high SD outliers. Again, these numbers are all AFTER fees.
Unless every human has started multiple hedge funds (tens of billions of coin flippers!), it seems unlikely that the statement that "no hedge funds justify their fees" is defensible, in my opinion.