making sense of 100's of funds

Patrick,

That is an interesting paper, in this case I was asking myself the 
question, "if I know that today's Melbourne prices are correlated to 
yesterdays  NY prices, is there a possibility of taking advantage of 
that?"  That is, not so much from putting a portfolio together, but 
daily investment decisions.

cheers

Patrick Burns wrote:

paul sorenson wrote:

Probably not unsurprisingly, the correlation between the SP200 and 
the global share fund daily returns goes up (from 0.16 to 0.42) when 
I insert a 1 day lag.
 

Careful.

There are asynchrony issues with global data, which definitely affect
the correlation (it is too low). Presumably the global return series has
asynchrony issues all by itself.

The working paper for Burns, Engle and Mezrich (1998) is available
at http://www.econ.ucsd.edu/papers/dp97.html

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.

http://www.metrak.com/tmp/exch10.png has plots produced by ccf().

Brian - if I want to look at returns over a different period using 
PerformanceAnalytics when the base data is daily, is the normal 
strategy just to undersample prices before calling CalculateReturns?

I am guessing that the correlation of the SP200 and global share 
fund would increase when looking at a longer time period.  I also 
want to compare it with some other data which comes out monthly.

cheers

BTW sorry if I am boring you guys with what must be very basic stuff 
- just tell me.
 

Asynchrony is neither basic (i.e., well-studied) nor very well 
appreciated.
I don't vote this message boring.

Patrick Burns
patrick at burns-stat.com
+44 (0)20 8525 0696
http://www.burns-stat.com
(home of S Poetry and "A Guide for the Unwilling S User")

Brian G. Peterson wrote:
 

paul sorenson wrote:
 

I ran a pairs plot on the daily fund returns as well as 
calculating the correlation coefficient (Pearson).

The pairs plot is reproduced at 
http://www.metrak.com/tmp/exch09.png and unless I am missing 
something, some of these "look" significant whereas some don't.
    

The pairs plot will certainly show you funds that closely track the 
index.  A quick check of cor() (or the Pearson correlation 
coefficient) and CAPM.alpha() will do the same.  A pairs plot (and 
to a lesser extent correlation) won't show you anything about 
systematic out-performance or under-performance, while alpha is a 
good indicator if the benchmark you choose is indicative of the 
investment universe of the fund.  Another good indicator is 
Sortino's Upside Potential Ratio, especially if you choose the 
benchmark index standard deviation as your MAR.

CAPM alpha will not be a good indicator if you choose an index that 
is different from the investment style of the fund.  For example, 
using a SP200 index with a fixed income fund wouldn't make any sense.

Cheers,

  -Brian