Time-weighted return: measuring manager skill
Time-weighted return measures the growth of one unit of currency invested at the start of the period, stripped of any distortion from client-driven cash flows. If a client adds a large deposit right before a market decline, TWR does not penalize the manager for that cash flow timing decision the manager did not control.
TWR is calculated by dividing the measurement period into sub-periods, each separated by a cash flow event, computing a return for each sub-period, and geometrically linking the sub-period returns. The geometric linking ensures that a 10% gain followed by a 10% loss produces a compound return of approximately negative 1%, not zero.
The Global Investment Performance Standards (GIPS) require TWR for composite performance presentations because institutional managers typically cannot control when clients contribute or withdraw capital. TWR allows performance to be compared across managers and time periods on a consistent basis.
TWR does not require daily valuation if a modified Dietz or modified IRR approximation is used between cash flow dates. However, large cash flows occurring on days with significant market moves can make approximations materially inaccurate. High-quality attribution requires daily valuation or at minimum sub-period valuation at every cash flow date.
Money-weighted return: measuring the investor experience
Money-weighted return is the internal rate of return (IRR) of the actual cash flows into and out of the portfolio. It is the discount rate that makes the present value of outflows equal to the present value of inflows plus the ending value. Unlike TWR, MWR is directly affected by the size and timing of capital contributions and withdrawals.
MWR is more appropriate when the portfolio manager does control cash flows: private equity funds, separately managed accounts where the manager has discretion over capital calls and distributions, and personal portfolio analysis where the investor wants to understand their actual dollar-weighted outcome.
A manager with strong TWR can have poor MWR if clients added capital before underperforming periods and withdrew before strong ones. Conversely, a manager with mediocre TWR may appear to have strong MWR if clients happened to contribute capital before strong performance. Neither number is wrong; they report on different aspects of the relationship between manager decisions and client outcomes.
For attribution purposes, TWR is almost always the correct choice. Attribution decomposes returns into allocation effects, selection effects, and interaction effects -- all of which should reflect manager decisions, not client behavior. MWR-based attribution conflates the two.
Return calculation errors that propagate into attribution
Incorrect corporate action treatment is the most common source of return calculation error. Stock splits, spin-offs, special dividends, and rights issues all require specific adjustments to the portfolio valuation and return calculation. An unadjusted price series treats a 2-for-1 split as a 50% price decline.
Income accrual timing errors occur when dividends, interest, or coupons are recorded on payment date rather than ex-date. This creates a one-time distortion in the period that spans the ex-date to the payment date and can be material for fixed income portfolios with many securities accruing income simultaneously.
Currency conversion errors affect multi-currency portfolios when local returns are translated to the reporting currency at different rates than the benchmark uses. Even small rate differences compound across many positions.
Benchmark return sourcing errors occur when the benchmark return used in attribution comes from a different data source than the official index administrator. Return differences between sources can reach several basis points per quarter -- which, at the security level, can shift an allocation or selection effect from positive to negative.
Frequently asked questions
What is time-weighted return in portfolio performance measurement?
Time-weighted return (TWR) measures the growth of one currency unit invested at the start of a period, eliminating distortion from cash flows that the manager did not control. It is calculated by linking sub-period returns geometrically across each cash flow event. TWR is the standard for comparing managers because it isolates investment decisions from investor deposit and withdrawal timing.
What is money-weighted return and when should it be used?
Money-weighted return (MWR) is the internal rate of return of the actual cash flows into and out of the portfolio. It measures the actual investor experience, including the dollar-weighted impact of timing contributions and withdrawals. MWR is appropriate when the manager controls cash flow timing -- private equity funds, certain managed accounts -- or when the investor wants to understand their real-world outcome rather than evaluate the manager's skill independent of cash flows.
When should you use time-weighted return versus money-weighted return?
Use TWR when evaluating manager skill, comparing managers, or presenting composite performance. TWR removes the effect of cash flows the manager cannot control, making fair comparison possible. Use MWR when the manager controls cash flows (private equity, managed accounts with discretionary capital calls) or when the investor wants to understand their actual dollar-weighted outcome. Using MWR to evaluate a public equity manager who cannot control client deposits is inappropriate because it penalizes or rewards the manager for investor behavior.
What are common return measurement errors in performance attribution?
The most common errors are: (1) Incorrect corporate action adjustments -- stock splits, spin-offs, and special dividends not reflected in the return series. (2) Income timing errors -- dividends or coupons recorded on payment date rather than ex-date. (3) Currency conversion inconsistencies -- local returns translated at different FX rates than the benchmark. (4) Benchmark return sourcing errors -- using an unofficial data source for the benchmark that differs from the index administrator's official returns. Each error propagates into every attribution output derived from the return.