Question about contribution to return analysis

I’m wondering if you can run contribution to return analysis over a 10-year period if your portfolio constituents have changed over that time. My challenge is that I basically have to deconstruct the composite performance on a monthly basis (manager request), and then funnel that information up over 10 years. Can’t really do a traditional Attribution model (selection/allocation/interaction effects) because we don’t have a policy portfolio to measure by (for the asset class)-- although I could just equal-weight I suppose.

I’m getting stuck beyond a year because of the annualization/compounding effects of the 10-year return. Can’t fit the monthly contributions nicely into the top-line number.

Anyone have experience with this that can suggest a solution?

Get FactSet