Most bond fund risk numbers come from a partial portfolio

Aggregate risk numbers reported for a bond fund — duration, yield, credit quality — arrive with an unquoted confidence interval, and disclosure of that interval stays rare. The figure is a weighted average computed from the positions where inputs exist. For many funds, that denominator falls well short of 100 percent of the portfolio. The remainder — private placements, structured instruments, derivatives lacking a clean price, securities missing reference data — contributes to the fund's actual risk while staying outside the reported figure. A duration of 4.2 years at 98 percent coverage and a duration of 4.2 years at 72 percent coverage make different claims, even though they look identical on screen.

The pattern runs through yield to maturity, average credit quality, and any sector or maturity weight built from position-level inputs. All are computed from a subset. The live questions: how large the subset runs, and whether the reader knows.

How the aggregation works

A fund-level duration comes from taking a bond position's modified duration, weighting by the position's share of the portfolio, and summing across the book. The inputs are the bond's coupon, maturity, yield, and price. Where an input goes missing — a private-credit position lacking a market price, a structured instrument with a complex cash-flow schedule, an interest-rate swap whose duration requires a model the aggregator skips — the position leaves the weighted average entirely. The denominator shrinks, and the surviving weights implicitly rescale to fill the gap.

The methodology is standard and defensible. The arithmetic stays clean. What it smuggles in is an implicit assumption: that the uncovered portion of the portfolio matches the duration of the covered portion. If the uncovered positions are short-duration cash equivalents, the reported figure overstates duration. If they are long-duration illiquid credit, the figure understates the risk. The reported number is correct for the slice it measured. Whether that slice represents the whole fund is a question the figure leaves open.

Coverage varies widely and rarely appears

Across the fund universe, coverage percentages for bond analytics range from near-total to well below three-quarters. A large-cap investment-grade fund built on Treasury and agency securities typically clears 95 percent coverage — nearly all its positions come with the inputs duration and yield require. A multi-sector or unconstrained fund with a mix of high-yield corporates, bank loans, CLO tranches, and emerging-market sovereigns may run coverage in the 70-85 percent range, because a meaningful share of the book consists of positions lacking clean pricing, standardized identifiers, or the reference data analytics require. Coverage runs independent of fund size, rating, and declared investment objective in any simple way. Coverage is a function of portfolio composition — specifically, the share of the book in instrument types where position-level analytics compute reliably.

The figure below is the pattern in delivered data. Each line is one of the largest bond funds registered with the SEC; each column is one measure of its bond sleeve. The length of a bar is the figure itself. The ink is the denominator: the share of the sleeve, by weight, that had the input the figure needs.

Index funds built on Treasury, agency, and investment-grade corporate paper print in full ink across every column: a price, a rating, and a maturity exist for nearly every position they hold. Funds holding municipal paper, private placements, or derivatives fade, and they fade unevenly — a rating or a duration is the first input to go missing, while price and maturity stay solid, because a stated maturity date and a mark are the two inputs almost every bond has. Two funds can print the same duration in the same column and rest on denominators a factor of ten apart.

The practical problem: the coverage percentage rarely appears alongside the headline figure. A data provider reports duration: 4.2 years. The feed rarely adds "duration: 4.2 years, computed from 78 percent of the portfolio by weight." The reader treats 4.2 as a fact about the fund. It is a fact about 78 percent of the fund, and the remaining 22 percent could push the true figure in either direction.

What SQXray prints instead

SQXray computes the bond-sleeve measures from the filed positions and the priced instruments, and it publishes the denominator beside each figure as a field of its own. In the portfolio analytics file every measure ships with a twin: modified_duration with modified_duration_coverage_pct, yield_to_maturity_pct with yield_to_maturity_coverage_pct, avg_credit_score with rating_coverage_pct, and the same pairing for coupon, maturity, price, convexity, and yield to worst. fixed_income_weight_pct states how much of the fund the sleeve is, fi_status states whether the sleeve was measured at all, and maturity_wal_pct states the share of the maturity figure that rests on a prepayment assumption rather than a stated date. Where the engine lacks the inputs for a measure, the figure is left blank and the coverage field beside it states the share that was measurable, so a blank reads as a measurement of the gap rather than as an omission.

The fund profile page prints the same pair on screen: every fixed-income figure sits beside a coverage chip naming the share of the sleeve it describes. Open the iShares Core U.S. Aggregate Bond ETF and the Vanguard Intermediate-Term Tax-Exempt Fund side by side and read the chips before the figures. The field definitions, with the methodology version each measure follows, are in the data dictionary.

The practical consequence

For a risk team using fund-level duration to size a hedge or set a VaR limit, the coverage gap wraps an unquoted confidence interval around the number. A fund at duration 4.2 with 98 percent coverage makes a different claim from a fund at duration 4.2 with 72 percent coverage, even though the two figures look identical on screen. The first fund's true portfolio duration falls almost certainly between 4.0 and 4.4. The second fund's true duration could plausibly range from 3.5 to 5.0 or wider, depending on the character of the uncovered sleeve.

Credit quality averages inherit the problem and add complications. The standard approach — mapping letter ratings to a numeric scale, averaging, and mapping back — is sensitive to the mapping function and obscures the distribution. A fund with an average credit quality of A could span AA and BB (a barbell) or cluster tightly in A (a concentration). The average marks the center of gravity. Whether the mass spreads or clusters stays hidden, and the positions lacking a rating stay outside the calculation entirely.

The coverage percentage — the denominator behind the weighted average — belongs beside the aggregate bond fund risk measures as a first-class field rather than a footnote, printed in the data feed and the risk report beside duration, yield, and credit quality. Lacking that field, identical numbers mean different things across funds, and readers lack any way to tell which. As discussed in Where fund disclosure rules stop, the disclosure framework establishes what must be reported but leaves ample room for a reader to mistake a partial figure for the whole. Coverage percentage is the instance of that pattern that matters most for fixed-income risk — and the most straightforward to fix.

Open a bond fund profile

Notes

1. The figure draws from the fund portfolio analytics feed for the analysis date stated in its caption. Each measure's line in the feed states the methodology version it follows (fixed_income_methodology_version) and the engine version that computed it (fixed_income_engine_version).
2. The funds shown are the largest SEC-registered bond funds by net assets on that date, taking each fund's largest share class with an ISIN. Money-market vehicles are excluded, and a fund qualifies when at least half its book is in bond-shaped positions (fixed_income_weight_pct).
3. Coverage is the share of the bond sleeve, by weight, whose positions had the input a measure needs: a price for the average price, a modified duration for duration, an agency rating or an equivalent for credit quality. For average maturity, maturity_wal_pct states the further share of the figure that rests on a prepayment assumption for amortizing pools rather than on a stated maturity date.
4. Field definitions are in the data dictionary; the worked figures in the prose (4.2 years at 98 and 72 percent) are illustrative, and the figures in the chart are delivered values.

Related reading

• Where fund disclosure rules stop What the disclosure framework requires a fund to report, and the room it leaves for a partial figure to read as the whole.
• How stale is a fund's publicly disclosed portfolio The lag between a portfolio as held and the portfolio the public record shows, measured across the filing calendar.
• Why share classes within a fund diverge in risk One portfolio, several share classes, and the mechanisms that make their risk differ.