[R-SIG-Finance] Using Benford's law for reported returns?

[Peter Carl]

I was recently inspired by a post on Drew Conway's blog:

http://www.drewconway.com/zia/?p=2234

to take a look at using Benford's law to look at returns as reported by
hedge funds.  I've made some headway:

https://r-forge.r-project.org/scm/viewvc.php/pkg/PerformanceAnalytics/sandbox/Benford.R?view=markup&root=returnanalytics

Currently this code can be used to calculate the theoretical proportions
and digit counts/percentages for first and n-th digits tests.  But as I've
been digging through Nigrini's papers, I'm realizing that this topic is
something of a deep hole, and these tests I've implemented may have only
modest utility.  Other tests might be more useful such as first-two
digits, number duplication, rounding, and last-two digits.

I'm aware of code in VGAM and rattle packages for Benford's distribution,
but the functionality is somewhat limited to what I've done already.

Does anyone have additional code in this vein that they would be willing
to share, or would you be willing to collaborate to extend these
functions?  Or, if you are a user of such analysis, I would be interested
to know if you've found it useful.

Thanks in advance,

pcc
-- 
Peter Carl
http://www.braverock.com/~peter