Hello,I'm trying to come up with an annual performance attribution for a fund of funds and I'm having trouble coming up with a method to do it. The data that I have is the following: monthly P&L for the fund, monthly beginning NAV, and monthly P&L broken out for each segment that I want attribution for (basically, each manager that the fund invests with). From this I can calculate the compounded annual rate of return for the fund, but I'm not sure how to break that number out so that I can say, for example, in 2006, the fund returned 20% and -5% came from manager 1 and 25% from manager 2. Here's a link to a google spreadsheet where put in some basic data as an exercise. http://spreadsheets.google.com/ccc?key= ... l=enColumn B is the NAV at the beginning of each month. In this example, I assume there are capital outflows for the first 11 months such that nav is kept at $1MM, and then there is an inflow of ~$99MM in the last month. Cell G14 is the annual return for the fund (18.09%). Cells E15 and C15 are compounded portions of the monthly fund return attributable to each manager. As one would expect, they don't add up to 18.09%. The problem is I can't think of another way to get the attribution that I need. The only thing I could think of, which I am sure is wrong, is to proportionally decrease those numbers until they add up to 18.09% which is what cell G16 is. Any thoughts would be much appreciated. Also, feel free to modify the spreadsheet if you want.

A nice way to check predictive ability is the AP decomposition by Andrew Lo.Best

as you say, you wouldn't expect the returns attributable to the managers to add up to the total refund in a time-weighted framework.My advice would be to think of a name for the fudge-factor (the difference between the two sets of numbers) and include that in the table

you're looking for RAPM, there's a bunch of "standard" techniques