IRR vs TWR: How an 8% Portfolio Return Became an Investor Loss

In the one-year example below, a portfolio has a +8% time-weighted return (TWR) while the family’s money-weighted IRR is about −1.33%. Its ending mark is US$20,000 below the family’s contributions. Both calculations use the same portfolio record; they answer different questions about the timing of the family’s money.

For a family office, the first question is therefore not “Which number is better?” It is what decision the number must support. Use a properly calculated TWR to examine a manager’s investment result without the family’s contribution timing. Use a money-weighted IRR to examine the family’s experience on its actual dated capital, while checking who controlled the cash flows and how any unsold assets were valued.

The worked case below uses invented values and no fees, income, distributions or taxes. It illustrates a method, not a market return or any family’s actual performance.

01What IRR and TWR Actually Measure

A TWR answers: How did the invested portfolio perform across the period, abstracting from the timing and size of external money? Divide the period at each external flow, calculate the return for each subperiod, then compound those returns. The measurement needs a reliable value just before or after each flow, with the flow handled consistently.

An IRR, also called a money-weighted return in this context, answers: What annualized discount rate balances the investor’s dated contributions, withdrawals and ending value? A larger contribution made just before a weak period has more influence on IRR than a small contribution made earlier. A terminal valuation is treated as an ending value in the calculation; it is not a cash distribution or realized profit.

CFA Institute’s 2026 overview of the GIPS standards says time-weighted returns are generally required for portfolios under those standards, with defined exceptions for money-weighted presentation. This is a standard for firms claiming GIPS compliance, not a rule that every family-office report must show only TWR. The metric still has to fit the question and the portfolio.

02One Portfolio, Two Returns: The Cash-Flow Test

Assume a family starts a one-year public-market account with US$1 million. During the first half, its value rises 20% to US$1.2 million. The family then contributes another US$1 million at the exact midpoint. The post-flow account value is US$2.2 million. During the second half, the portfolio falls 10%, ending at US$1.98 million.

View the full-width desktop return timeline. The diagram uses equal half-year periods and no fees or taxes.

The TWR links the two portfolio returns: (1 + 20%) × (1 − 10%) − 1 = +8% for the full year. The midpoint contribution changes the dollars exposed to the second half, but not either subperiod’s percentage return.

For investor-level IRR, use the family’s perspective: −US$1 million at the start, −US$1 million at the midpoint, and +US$1.98 million as the ending value. With exactly equal half-year intervals, solve 0 = −1m − 1m/(1+r)^0.5 + 1.98m/(1+r). The effective annual IRR is approximately −1.33%. The family contributed US$2 million and ends with a marked account worth US$1.98 million: a US$20,000 mark-to-market loss before fees and taxes.

These answers coexist because the family placed another US$1 million into the portfolio before the −10% half. The +8% is the compounded return across two portfolio subperiods; it is not 8% earned on the family's US$2 million of dated contributions. The second US$1 million was exposed only to the falling half.

TWR describes the sequence of portfolio percentage returns. IRR describes the family’s dated capital in that same sequence. Neither tells the whole story alone, and the negative IRR is not proof that the manager caused the timing of the family’s contribution.

The one-year case avoids a common comparison error: do not put a one-year annualized IRR next to a six-month cumulative TWR without stating the periods. If actual flows fall on calendar dates rather than an exact midpoint, use those dates; a date-based IRR may differ slightly from this equal-half-year illustration.

03Which Return Belongs on Which Decision?

For manager evaluation, ask for a TWR calculated on the manager’s defined portfolio and compared with a relevant benchmark over the same period, with compatible fee treatment. This helps isolate investment performance when the family, rather than the manager, controls the size and timing of contributions. It does not excuse a manager from choices the mandate actually gave them, nor does it make an unsuitable benchmark useful.

For the family’s capital experience, ask for a money-weighted return on the family’s dated cash flows. It shows what happened to the dollars committed to that account or investment under the chosen end value. Pair it with the dollar gain or loss and with a record of who chose contribution dates. A family-initiated transfer, a manager-called capital draw and a distribution are not the same decision.

For asset allocation across a whole office, neither metric is a standalone verdict. An account TWR, an investor IRR on a private fund and a total-portfolio return may have different perimeters, cash-flow conventions, ending valuations and fee layers. Align period, perimeter and fee basis for a like-for-like comparison; fund-level and investor-level figures can legitimately differ because they describe different capital. Do not average unrelated IRRs or infer a consolidated family return by adding them.

04Why Private-Asset IRR Needs Its Own Questions

Private funds commonly report an IRR because capital is called and distributed across irregular dates; Commonfund's TWR/IRR explanation describes this common reporting distinction. That does not mean “private assets use IRR, public assets use TWR” is an absolute rule.

Under the GIPS framework, CFA Institute describes the money-weighted exception as requiring firm control over external cash flows and at least one specified portfolio condition such as closed-end structure, fixed life, fixed commitment or significant illiquidity. The standard’s conditions should not be collapsed into a shortcut for every private holding.

Ask whether a stated IRR is gross or net, fund-level or this family’s investor-level, and realized or dependent on an ending net asset value. A family may receive capital calls on different dates from another investor; its own return need not equal a fund’s headline figure. A fund valuation that has not been updated can make an interim IRR look precise while the underlying asset remains uncertain.

The critical inputs are few, but they need owners. Request a dated flow ledger showing capital calls, distributions and fees; the valuation date and source for the remaining holding; the measurement period; and the calculation method. Treat an undrawn commitment as a future obligation, not as money already invested or as a realized loss.

If a subscription facility delays a capital call, ask how the reported IRR treats that timing before comparing funds. ILPA's subscription-line guidance explains that delayed calls can enhance reported IRR and complicate comparisons. Its 2017 guidance explains a mechanism; it is not a current return estimate for any fund.

05What to Check Before Trusting Either Number

A return calculation should be reproducible from a small package of records. The checklist is especially useful when an office combines custodian data, private-fund statements and entity books in one family report. For wider source-to-report controls, see seven warning signs in family-office reporting.

Run the simple case above as a control. If a system reports +8% TWR but cannot show the +20% and −10% subperiods or treats the midpoint US$1 million as investment profit, hold the result. If it reports an IRR without the dated family flows and ending mark, request those inputs before using the figure in a decision. A displayed percentage is an output, not its own evidence.

06Put Both Returns on a Decision-Ready Report

For each mandate, place TWR, money-weighted IRR where useful, dollar gain or loss, and the comparison period beside one another. Label the portfolio or investor whose return is being shown. Add the external-flow dates, valuation source, fee basis and benchmark. Mark a missing input unknown, not zero.

The hypothetical report can put the numbers and unresolved evidence in one compact record:

The ending value is a mark, not a sale. A real office must populate this template and recalculate both returns from its own dated flows and verified values before using them in a decision. It can then discuss the manager-performance and family-capital questions separately.

Use a result only after the preparer can trace it back to values and flows and the reviewer has cleared material exceptions. If the report mixes dates, fee bases or owners, the right next step is a corrected return package—not a confident choice between IRR and TWR.