Skip to content
Learning path Gold Professional operator

Performance benchmark

A reference selected in advance to evaluate a portfolio's return and risk, defined with a comparable universe, total-return variant, currency, hedge and methodology.

Who this is for — Anyone deciding whether a portfolio's result was consistent with its mandate, while avoiding references selected after the outcome or series built on incompatible bases.

A performance benchmark is a reference used to evaluate a portfolio's return, risk and decisions. It may be a published index, a market portfolio, a cash rate, a blend of indices or a custom portfolio. An index does not become a benchmark automatically; it becomes one when it is selected to represent the mandate or evaluation context.

Selection should be ex ante, documented before the period or decision to be judged. Choosing afterwards whichever index makes the result look best introduces selection bias. A familiar benchmark can still be inappropriate when its universe, risk, currency, hedging policy, distributions or constraints do not match the portfolio.

A useful benchmark is chosen before measurement It should represent the mandate and make the comparison verifiable A useful benchmark is chosen before measurement It should represent the mandate and make the comparison verifiable mandate → specified benchmark → active return → attribution 1 Appropriate Consistent with the mandate’suniverse, style, currency and risk. 2 Measurable Prices, weights, income and calendarare observable and reproducible. 3 Unambiguous Constituents and rules areidentifiable without hindsightchoices. 4 Investable Represents a realistic passivealternative for the stated purpose. 5 Specified in advance The choice precedes the evaluationperiod. 6 Governed Methodology, changes andaccountability are documented. Cyclepedia · educational diagram: state conventions, period and data
A sound comparison depends on the reference's complete identity, not the familiarity of its name. Every uncontrolled difference can change active return and derived metrics.

A benchmark is a specification, not a ticker

Field Question that must be explicit
Universe and purpose Which instruments, region, style, maturity or risk profile does it represent?
Construction rule How are constituents selected, weighted, rebalanced and replaced?
Return variant Price return, gross total return or net total return?
Currency Local currency, base currency or another publication currency?
Currency hedge Unhedged or hedged; with what ratio, frequency and method?
Calendar and timing Which closes, holidays, time zones, cut-offs and valuation dates?
Cost and tax Which withholding assumptions, fees or costs are embedded?
Source and version Which administrator, exact identifier, methodology and effective date?

Professional guidance summarised by CFA Institute uses properties such as clarity, measurability, appropriateness, investability and advance specification. A benchmark need not reproduce every position, but structural differences should be relevant to the mandate and disclosed. A benchmark-agnostic portfolio may not be managed to minimise index deviation; that does not permit a convenient reference to be selected after the fact.


Price, gross total and net total return

A price return index measures price changes under the provider's rules and excludes ordinary distributions. A gross total return index reinvests gross distributions. A net total return index reinvests distributions after theoretical withholding taxes specified by the provider.

“Gross” and “net” in an index name often refer to dividend withholding, not automatically to portfolio returns before or after management fees, transaction costs or an investor's actual tax. MSCI and S&P Dow Jones Indices document price, gross-total and net-total families separately; the exact code matters.

Comparing a dividend-receiving portfolio with a price-only index can attribute to management a return that arises only because the reference excludes the distribution. Within the GIPS framework for firms, an appropriate total-return benchmark is used when available, while price-only information can be labelled as supplemental. That is a rule within the GIPS scope, not a universal law for every financial communication.


Local currency, base currency and hedging

The same basket produces different series in different currencies. A local-currency return seeks to reflect component markets under the index methodology. An unhedged return in the investor's base currency also contains currency movement. A currency-hedged index adds theoretical FX contracts, hedge ratios, roll dates, rates and mismatch between estimated and actual exposure.

A hedged benchmark is not merely a translated local series: its hedge method creates its own return and cost and can be over- or under-hedged between resets. A euro portfolio compared with a dollar index may show active return that is mostly an uncontrolled currency mismatch.


Making portfolio and reference comparable

Comparison requires the same period, valuation timing and frequency; compatible total-return conventions; coherent currencies and hedge states; disclosed fees, transaction costs and tax; compatible cash and financing treatment; and consistent treatment of corporate actions and missing data.

Investability is contextual. A market index can be a useful opportunity-set reference even though no investor can hold it without frictions. The gap must be understood: turnover, capacity, taxes and replication costs may make the published series unattainable.


Composite and custom benchmarks

A multi-asset mandate may use a blend such as several indices plus cash. Its rules need component weights, rebalancing dates, return variants, currencies, hedging and treatment of a component that stops publishing. Changing a blend is a policy event with an effective date, not a silent history rewrite.

A custom benchmark can represent a specialised mandate better than a broad index, but it introduces additional governance. Constituent selection and weights must be known in advance, the series must be calculable and records must preserve every methodology version. An unreproducible “peer average” or target assembled after the period is not a neutral benchmark.


Metrics inherit the benchmark choice

For period t, arithmetic active return is:

aₜ = Rₚ,ₜ − Rᵦ,ₜ

Tracking error is the standard deviation of periodic active returns, and the information ratio divides their mean by that dispersion:

TE = sd(aₜ);   IR = mean(aₜ) / TE

If TE is zero, IR is undefined. Adding periodic active returns is not always the same as subtracting cumulative returns because compounding must be handled consistently. Beta, alpha and attribution also inherit the benchmark and data model. A misspecified reference changes the question, not just the noise.

There is no universal tracking-error or information-ratio threshold that makes a strategy good or suitable. A low TE may fit an index mandate or conceal cost without value; a high TE may be intentional or an unwanted breach. Mandate, horizon, capacity, cost and uncertainty come before a label.


Invalid comparison and correction

Suppose a global portfolio reports +7% net of fees in euros and is compared with a +5% price-return version of an index in dollars. Claiming “+2% outperformance” is unsupported because distributions, currency and fee basis are not aligned.

In the same hypothetical example, the unhedged net-total-return version in euros might be +9%, making comparable active return 7% − 9% = −2%. If the mandate required currency hedging, the corresponding hedged series would be needed instead. The numbers illustrate how the benchmark identity can reverse a conclusion without changing the portfolio.


Governance checklist

IOSCO's benchmark principles address governance, quality, methodology and accountability. GIPS governs compliant performance presentations by firms that claim compliance. Both have specific scopes and do not replace applicable law, prospectuses, contracts or internal policy.

A reproducible process preserves the ex-ante choice and rationale; exact identifier and methodology; price/gross/net variant, currency and hedge state; calendar and data source; fees, cost, tax, cash and financing treatment; blend and rebalancing rules; dated changes; and structural differences. Alternative benchmarks may be diagnostic when clearly labelled, not retroactive substitutes.

Common error — Using the S&P 500, zero return or the best index found afterwards as a universal reference. The benchmark must represent the mandate in a fully identified variant; it is neither investment advice nor a guarantee of replicability or return.


Sources