Actuarial Outpost (http://www.actuarialoutpost.com/actuarial_discussion_forum/index.php)
-   Finance - Investments (http://www.actuarialoutpost.com/actuarial_discussion_forum/forumdisplay.php?f=12)
-   -   Testing manager performance i.e. comparing two samples (http://www.actuarialoutpost.com/actuarial_discussion_forum/showthread.php?t=334290)

 Kenny 08-02-2018 04:49 PM

Testing manager performance i.e. comparing two samples

I have (for example) a data set of annual investment returns over a period of 10 years for 100 investment managers. They are ranked in some manner from 1 - 100. I think that managers ranked from 11-20 achieve better returns on average than managers ranked from 81-90 so I want to run a statistical analysis comparing these 2 groups, but I am not completely clear on the proper way to test this hypothesis.

I am looking at using a t test comparing the means of the 2 samples where the variance of the populations are assumed to be equal. Should I be running this test using:

1) average annual returns of each group over the 10 year period such each sample has 10 observations

2) actual returns of each manager over the 10 year period such that each sample has 100 observations

3) actual returns of each manager in a given year and aggregate the results across time somehow

4) ?

What if ranking is based on assets under management? Given that over a 10 year period the groupings may change slightly does that impact the way in which the analysis should be performed?

 Kenny 08-03-2018 01:31 PM

Reposted, not moved, to non-actuarial

http://www.actuarialoutpost.com/actu....php?p=9399218

I'll delete this thread if I get traction there but not here.

 ElDucky 08-03-2018 01:35 PM

a t or F test probably. I didn't read the whole question.

 Kenny 08-03-2018 03:03 PM

Quote:
 Originally Posted by ElDucky (Post 9399233) a t or F test probably. I didn't read the whole question.
Thanks. That sounds about right for a Friday afternoon. :toast:

 All times are GMT -4. The time now is 12:19 PM.