Calculate rolling performance in R

Question

I have the following dataframe,

           FECHA     Pxlast    
1      2010-12-31    332.636        
2      2011-01-07    334.327        
3      2011-01-14    341.771        
4      2011-01-21    331.241        
5      2011-01-28    333.252 

I have to calculate a new column called "Rolling 4 weeks", the values are based on the following idea, for example for index 5, it will be pxlast[5]/pxlast[5-number of weeks]-1.

Then I get the performance for 4 weeks, in this example it should be pxlast[5] = 333.252 , pxlast[5-4] = 332.636, then I divide it and I subtract - 1 , so the result is -0,384.

Ok, I can do it using "for" loop ; but reading about some functions that could do it properly I find the function chart.rollingPerformance, from the PerformanceAnalytics.

It applied the rolling over a parameter called FUN, for example "mean", it will calculate the mean between width space, but I don't know how to calculate Performance correctly.

Here is the output dataframe for example.

       FECHA           Pxlast          Rolling4W
1      2010-12-31    332.636               NA
2      2011-01-07    334.327               NA
3      2011-01-14    341.771               NA
4      2011-01-21    331.241               NA
5      2011-01-28    333.252              -0,384

The NA values are because we are calculating performance from a width of 4 spaces ( weeks ).

Is there any function to do it without loops?

Answer 1

require(data.table)
d <- data.table(x = 1:5, y=c(1, 2, 4, 7, 11))
d[, z := shift(y, 4)]
d[, z := y/z - 1]
d
#    x  y  z
# 1: 1  1 NA
# 2: 2  2 NA
# 3: 3  4 NA
# 4: 4  7 NA
# 5: 5 11 10