0
$\begingroup$

i am building a platform for portfolio analytics, part of which is a performance attribution module. Given that most individual portfolio's can never have the same number of stocks as say, a mutual fund, i was wondering how do i perform a stock weighting attribution for a portfolio. For instance if my portfolio had 6 stocks from 3 sectors (Energy , Utilities and Telecom) which have a split of 30-40-30 in my portfolio how do i compare it to the market weightage? These sectors have a weight of 8%, 3% and 2.4% in the S&P 500 ..i don't get anything by comparing 30-40-30 with 8-3-2.4 instantly ..but if i normalize 8-3-2.4 into 60-22-18 i can tell that my pf did better / worse because i was over/under weight one of the sectors. Can i use a normalized method for comparison ? most online sources and papers give a plain vanila approach assuming this can only be used by fund managers et al .. kindly correct me if i am wrong.

Regards, vikram

$\endgroup$
  • $\begingroup$ I would recommend reading through the two papers given in the second answer to this question: quant.stackexchange.com/questions/2903/… $\endgroup$ – Dr_Be Feb 10 '16 at 14:39
  • $\begingroup$ Hi BerndH ..thanks for pointing the papers out but i had already read the morningstar paper ..the second one follows pretty much the same track, but both these papers assume a portfolio with a lot of sectors / diversity ..my problem is if i have half as many or lesser .. i am pretty sure we can still benefit from PA but the methodology is what is keeping me a little twisted $\endgroup$ – Vikram Murthy Feb 10 '16 at 15:10
  • $\begingroup$ OK, I see your point now. Apart from the pure technique (applying Brinson-like methods) your main question is: What is my benchmark? From what you've written it is apparently not the S&P500. You can restrict it to the three sectors (or maybe some more if you have chosen an active weight of 0 % for them in your portfolio). In this case your bm would be a linear combination of the respective S&P500 sector indices with the "normalized" weights. But your free in the first place and can also apply 33.33% to each of them. $\endgroup$ – Dr_Be Feb 10 '16 at 16:01
  • $\begingroup$ Thanks so much for responding BerndH .. yeah, let me give that a shot and see how the users interpret the comparisons :) ..thanks again $\endgroup$ – Vikram Murthy Feb 11 '16 at 2:23
1
$\begingroup$

Effective PA is dependent on the correct description of the investment process. I am not sure, from what you say, what exactly is your investment process. But let me presume that it is the following: You have chosen the S&P500 as your benchmark. You first distributed your money among sectors. (That you gave many sectors zero weight is not relevant to the methodology, only to the resulting evaluations.) Then you distributed your money within each of the Sectors, making one decision for each sector. In sectors that you put no money, it is assumed that the zero value was distributed as the BM did. Thus, a PA model should rightly only evaluate your four decisions. A straightforward application of a single period PA would provide these results. Of course, if there were any trading during the period, you would need a multi-period PA in order to fairly evaluate your decisions.

$\endgroup$
  • $\begingroup$ thanks so much , you pretty much nailed it (the investment process) ..so to be precise i am trying to build a platform for a few friends that regularly invest but don't exactly invest in all the sectors. Hence the confusion. But with the statement "..In sectors that you put no money, it is assumed that the zero value was distributed as the BM did.", you cleared any remaining confusion i had..so deeply appreciated. Sadly i don't have enough repo to upvote ..thanks again $\endgroup$ – Vikram Murthy Feb 16 '16 at 2:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.