Methodology — Modified Dietz Return

Transparency is a feature. This page explains what the Modified Dietz return measures and why we compute it — in plain terms. The precise formulas and numeric conventions are documented in our internal methodology; the summary here is deliberate, not an omission.

The Modified Dietz method estimates a money-weighted (dollar-weighted) return over a single measurement period. Its purpose is to measure how the money actually invested performed, while accounting for cash that moved in or out partway through the period. It does this by crediting each external cash flow only for the portion of the period it was actually at work — a fast, closed-form approximation to a true internal rate of return (IRR/XIRR) that avoids the iterative root-finding the exact method requires.

This method underpins PORT's transaction-aware linked attribution (see the linked Brinson methodology): each sector sleeve's monthly return over a saved book's ledger is a Modified Dietz return with that month's trades treated as external flows — so intra-month buys and trims move capital without masquerading as performance.

Modified Dietz return

This is the period's gain after removing the effect of money added or withdrawn, expressed relative to the average amount of capital actually employed during the period. The result is a single-period, unannualized figure read as a decimal fraction, where a positive number is a gain and a negative number is a loss. Cash added to the portfolio (a contribution) and cash taken out (a withdrawal) are treated with opposite signs, so neither one is mistaken for investment performance. When there are no external flows at all, the method simply collapses to the ordinary return: ending value versus beginning value.

Dietz time weight

The time weight is how much of the period a given cash flow was invested, expressed as a fraction ranging from fully invested down to not invested at all. A flow that arrives at the very start of the period is treated as fully invested for the whole period; one that arrives on the final day contributes essentially nothing to the period's earning capital; a flow halfway through counts for about half. These weights are what let the return above measure average capital employed rather than assuming every dollar was present the entire time.

Honest limits

• This is a single-period estimate, not annualized and not compounded.
• It is an approximation to the exact money-weighted (IRR) return; it is at its most accurate when flows are modest relative to the portfolio and reads best over shorter windows where the timing assumption holds up well.
• It needs a meaningful amount of capital at work to be defined — if the average capital employed nets out to nothing, the return is left undefined rather than reported as a misleading number.
• A zero-length period has no time for anything to be invested, so a time weight cannot be formed for it.