I am an R newbie, and I am facing what is for me a challenging problem. I am
trying to write functions to solve the following problems. It seems as though R
offers so many options I am not sure how to approach the problem.
I have a data frame with dates in the first column and returns in the remaining
columns. I would like to give a user a variety of options.
I designate Series1 as the base security. I want to find the correlation for
Series1 with every other security, but I want to do so:
1. By seasons and years
2. By seasons only
A third problem is a standard correlation matrix of all series but the output is
by season and year.
Typically the season is quarters. But it would be nice if the user could
designate the endpoints of each season.
In output 1 below, the correlations for 1Q2008 are computed by just using the
data that is within 2008-01-01 to 2008-03-31. In output 2, there are only four
rows of outputs, one for each quarter.
In output 3, the function would subset each quarter of each year and then use a
standard correlation.
The input
> head(ReturnsAll)[,1:4]
Dates Series1 Series2 Series3
1 2008-01-03 -0.002 -0.002 -0.002
2 2008-01-04 -0.008 -0.009 -0.009
3 2008-01-07 -0.025 -0.024 -0.024
4 2008-01-08 0.012 0.012 0.012
5 2008-01-09 -0.012 -0.014 -0.015
6 2008-01-10 -0.023 -0.022 -0.021
> tail(ReturnsAll)[,1:4]
Dates Series1 Series2 Series3
850 2011-05-18 0.021 0.021 0.021
851 2011-05-19 -0.008 -0.008 -0.008
852 2011-05-20 0.009 0.009 0.009
853 2011-05-23 -0.020 -0.020 -0.020
854 2011-05-24 0.022 0.022 0.021
855 2011-05-25 0.021 0.021 0.021
Here is how the output matrices would look.
Correlation between series 1 and all others
By quarter and by year
Date
Series1
Series2
Series3
1Q 2008
1.00
0.28
0.11
2Q 2008
1.00
-0.07
-0.07
3Q 2008
1.00
-0.08
-0.32
4Q 2008
1.00
0.16
-0.34
1Q 2009
1.00
0.09
0.04
Output 2
Correlation between series 1 and all others
By quarter only
Date
Series1
Series2
Series3
1Q
1.00
-0.06
0.58
2Q
1.00
0.25
0.48
3Q
1.00
0.52
0.59
4Q
1.00
0.54
0.11
Output 3
All correlations
By quarter and by year
Q1 2008
Series1
Series2
Series3
Series1
1.00
0.78
0.73
Series2
0.78
1.00
0.98
Series3
0.73
0.98
1.00
Q4 2010
Series1
Series2
Series3
Series1
1.00
0.97
0.95
Series2
0.97
1.00
1.00
Series3
0.95
1.00
1.00
Thanks.
Ira
[[alternative HTML version deleted]]