Serving the Quantitative Finance Community

 
User avatar
Manishs
Topic Author
Posts: 3
Joined: September 13th, 2002, 4:42 am

Tracking error

February 3rd, 2005, 11:08 am

Can someone provide me the formula for calculating the tracking error.
 
User avatar
Aaron
Posts: 4
Joined: July 23rd, 2001, 3:46 pm

Tracking error

February 4th, 2005, 2:34 am

Tracking error is a general term for differences between a portfolio return and a benchmark, or sometimes a derivative and a hedge position. It can be measured by standard deviation, mean absolute deviation or other statistics. The frequency of observation is also important.
 
User avatar
Manishs
Topic Author
Posts: 3
Joined: September 13th, 2002, 4:42 am

Tracking error

February 4th, 2005, 2:39 am

Thanks Aaron.Would the following description be correct.First the difference between returns on the benchmark, say, the Nifty index and the index fund is computed. The tracking error is the standard deviation of this difference. The tracking error is, therefore, the fluctuation in the differential returns between the index fund and its benchmark.
 
User avatar
tigerbill
Posts: 1
Joined: April 22nd, 2004, 7:14 pm

Tracking error

February 4th, 2005, 2:49 am

what you said is ex post tracking error.the formular for ex ante tracking error is different. you may refer to some book or simply search it online.
 
User avatar
Aaron
Posts: 4
Joined: July 23rd, 2001, 3:46 pm

Tracking error

February 4th, 2005, 1:49 pm

I agree except that there are other metrics for tracking error than standard deviation. Tracking error is the differences in return, although the name is also used for summary statistics about this difference. Standard deviation is the most common.As tigerbill says, ex ante tracking error is more complicated. It could refer to the distribution of the actual future tracking errors, or some statistics on that distribution.
 
User avatar
janickg
Posts: 0
Joined: August 3rd, 2004, 1:13 pm

Tracking error

February 10th, 2005, 7:41 pm

One way to calculate tracking error quickly...(wCurrent - wBench) * CovarianceMatrix * (wCurrent - wBench)'where wCurrent is a row vector of current weight allocations, and wBench is a row vector of benchmark weight allocations, CovarianceMatrix is needless to say your absolute covariance matrix.