# Performance evaluation - interesting find

Hey, Not sure if others have picked up on this but I was just re-reading SS 17 in Schwesser and realized that in the micro attribution section there is a Professor’s note that clarifies that combining the allocation/selection interaction with the within sector equation causes the bechmark weight to cancel out. So this ends up being equivalent to using the portfolio weight for determining security selection effect in global portfolio attribution. Somehow missed this until now…

right, global portfolio skips that step

Nice pick up. This has confused me since day dot. Do you mind setting out the math?

Can you eleborate on the details ? I am very curious.

Mathematically this is easy: Interaction: (Wp,j - Wb,j) (RP,j - RB,j) Within Sector: (Wb,j) (RP,j - RB,j) adding these together: (Wp,j) (RP,j - RB,j)

Thanks, sorry i still dont understand, you may need to take out the sledgehammer! Within-sector = performance attributed ONLY to security selection. (Rp-Rb)Wb Multiply the difference in return by the weight of the benchmark. Security selection = performance attributed to superior security selection. (Rp-Rb)Wp Multiply the difference in return by the weight of the portfolio. Question - why are these two formulas not the same