Comments on: Researchers to hedge fund investors: Don’t throw away Sharpe ratios just yet http://allaboutalpha.com/blog/2008/09/14/researchers-to-hedge-fund-investors-dont-throw-away-sharpe-ratios-just-yet/ Hedge funds, portable alpha, 130/30 and alpha-centric investing Fri, 19 Mar 2010 07:52:26 +0000 http://wordpress.org/?v=abc hourly 1 By: Bill aka NO DooDahs! http://allaboutalpha.com/blog/2008/09/14/researchers-to-hedge-fund-investors-dont-throw-away-sharpe-ratios-just-yet/comment-page-1/#comment-134093 Bill aka NO DooDahs! Mon, 15 Sep 2008 11:37:05 +0000 http://allaboutalpha.com/blog/?p=3395#comment-134093 While the methodology is straightforward, it is flawed. It measures merely the statistical tendency for high-ranking Sharpes to also rank highly on Sortino, etc. A histogram of how many funds changed rankings, and by how much, would be a better metric. For example, of the 1000s of funds, rank them by deciles using Sharpe. What percentage of funds changed decile categorization by no deciles, one decile, two deciles, etc., when re-ranked by each other metric? Of course, from the POV of a potential investor trying to determine whether Fund A is better than Fund B, no such "average-based" methodology provides help. If one has a minimum acceptable return, then one needs to use an evaluation metric that incorporates the minimum acceptable return. It really is as simple as that ... While the methodology is straightforward, it is flawed. It measures merely the statistical tendency for high-ranking Sharpes to also rank highly on Sortino, etc.

A histogram of how many funds changed rankings, and by how much, would be a better metric.

For example, of the 1000s of funds, rank them by deciles using Sharpe. What percentage of funds changed decile categorization by no deciles, one decile, two deciles, etc., when re-ranked by each other metric?

Of course, from the POV of a potential investor trying to determine whether Fund A is better than Fund B, no such “average-based” methodology provides help.

If one has a minimum acceptable return, then one needs to use an evaluation metric that incorporates the minimum acceptable return. It really is as simple as that …

]]>