Benchmark

A benchmark functions as a contract term, not a decoration. To serve that role it must be specified and agreed before the measurement period begins, otherwise the manager can select the yardstick after the fact and guarantee a flattering comparison. It must be unambiguous: the identities and weights of its constituents knowable at any point in time. It must be investable, meaning the fund could have simply held it as a passive alternative rather than paying for active management. It must be measurable at the required frequency, appropriate to the manager's stated investment approach, and reflective of current investment opinion, so the manager has an informed view on the securities inside it.

The final property is behavioural: the manager must accept the benchmark as their own. A yardstick the portfolio manager privately rejects will not discipline decisions, and disputes over attribution results will follow every reporting cycle.

Once a benchmark is fixed, performance becomes arithmetic. Active return for a period is the portfolio's total return minus the benchmark's return over the identical interval, using the same currency and the same treatment of income and fees. At the holdings level, that difference decomposes into active weights: for each security, the portfolio weight minus the benchmark weight, multiplied by the security's return, summed across the universe. A security absent from the portfolio but present in the index carries a negative active weight, so avoiding a stock is itself a position.

Active risk, commonly called tracking error, is the standard deviation of those periodic active returns, annualised. Dividing mean active return by active risk gives the information ratio, the standard measure of how efficiently a manager converts deviation from the benchmark into value added.

Several construction approaches coexist. Broad market indices are transparent and investable but frequently mismatched to a specialist mandate. Style indices narrow the comparison to value, growth, or a capitalisation band. Manager peer-group universes rank a fund against similar funds, but a median return is neither investable nor unambiguous and is contaminated by survivorship bias, so it fails several validity tests. Factor-model and returns-based benchmarks fit exposures statistically, while custom security-based benchmarks are built by sampling the manager's actual opportunity set and rebalancing on an agreed schedule.

Where the assigned benchmark differs from the manager's true investment universe, total active risk splits into misfit risk, arising from that mismatch, and true active risk, arising from genuine security selection and timing decisions.

Benchmarks appear in the investment policy statement, in composite reporting under global performance standards, and in fee schedules where performance fees accrue only above a stated index. They also drive risk budgeting: a mandate may cap tracking error, which constrains how far active weights can deviate from index weights. Index construction matters here, since capitalisation weighting concentrates exposure in the largest names and can quietly shift a benchmark's character over time.

A candidate should be able to compute active return from portfolio and benchmark returns, build tracking error and the information ratio from a series of active returns, judge a proposed benchmark against the validity properties, and explain why a peer-group median or a style-mismatched index distorts the assessment of a manager's skill.