Performance Measurement

Performance measurement asks what the return was; performance attribution asks where the return came from (and which risks produced it); performance appraisal asks whether the result reflects skill. Measurement feeds attribution, and attribution feeds appraisal.

It accounts for the entire portfolio return (or active return) with no large unexplained residual, reflects the manager's actual investment decision process, quantifies each active decision, and is computed from data and a benchmark that match the strategy.

Return attribution decomposes the active (or total) return into the effects of decisions such as allocation and selection. Risk attribution decomposes the portfolio's risk (volatility or tracking risk) into the contributions of those same decisions or factors.

Macro attribution is done at the fund-sponsor (asset owner) level: it measures the effect of the owner's decisions such as the strategic policy mix, tactical tilts and manager choices. Micro attribution is done at the individual manager's portfolio level, measuring that manager's allocation, selection and other decisions versus its benchmark.

It regresses the portfolio's total returns on a set of factor or index returns to infer average exposures and attribute returns. It is cheap and needs only a return series, so it suits funds without holdings transparency (e.g., many hedge funds), but it is the least accurate approach: it captures only average exposures over the estimation window and misses changes in positioning.

Holdings-based attribution uses beginning-of-period (or periodic) holdings and assumes they are held through the period, so it misses intraperiod trades and leaves a residual; accuracy improves with shorter periods. Transactions-based attribution uses both holdings and every transaction, so it fully explains the return, but it is the most data-intensive and costly.