making sense of 100's of funds

On 8/19/07, Patrick Burns <patrick at burns-stat.com> wrote:

That paper talks about the effects of asynchrony and proposes a
method of backing out data without asynchrony.  Our investigation
suggested that using weekly data is adequate for avoiding asynchrony
effects.

We have done a lot of work on this problem and agree with the choice
of weekly data.

Paul,

The use of benchmarks may not be the optimal path in this application,
relatively simple ranking might be more viable. As a compromise, you
might try looking at ranked Sharpe ratios...

A stack ranking of risk/reward ratios is a good idea.  I would recommend 
using either a Cornish Fisher modified Sharpe ratio (to take possible 
non-normality of distributions into account) or Sortino's Upside 
Potential Ratio.  Even Sharpe himself recommends the use of Information 
Ratio preferentially to the original Sharpe ratio, but old habits die 
hard...

To answer an earlier question on sub-sampling:  Yes, from daily *price* 
data you can construct a weekly or monthly series by simply taking the 
price at the end of the week or month, and constructing your returns 
series from the end of period closing price.  The zoo library also has a 
good implementation of the aggregate() function for timeseries data to 
help you automate the sampling while maintaining your original daily data.

Regards,

    - Brian