A performance measurement myth: rolling up is better

A common myth that I often run into is the one involving building up returns from the underlying assets. That is, the belief that if you calculate the returns at the security level and then "roll them up" to derive the portfolio return, that is somehow better.

WRONG!

It is better to calculate the portfolio return based on its beginning and ending market values, its cash flows, and ideally, its values when flows occur.

Building up from the securities can often create errors. Plus, as I've explained before, security level returns should be money-weighted, not time-weighted, thus their use for rolling up has to be wrong!

I just reviewed one client's system that calculated returns this way, and discovered that not only were the portfolio and asset class returns in error, so were the security returns themselves!

Let's put an end to this myth!