Methods

Fund-reported bond prices

US mutual funds and ETFs file SEC Form N-PORT with every holding: identifier, face value, value in dollars, and for bonds the coupon, maturity and default, arrears and payment-in-kind flags. The SEC publishes the third month of each fund's quarter about 60 days after month-end, and funds report on staggered quarter-ends, so most bonds get a mark every month.

Price per 100 of face value = value ÷ face value × 100, in the bond's own currency (funds' exchange rates are checked against each other; a rate reported upside down is inverted). Lots of the same bond in one fund are summed. Across funds, a mark more than five points and five median deviations from the other funds' marks is dropped; the price shown is the value-weighted average of the rest. A month counts as complete when it has at least 80% of the reports filed for the same month a year earlier; partial months are labelled and never feed changes in holder counts.

Holder changes compare each fund series with its own previous report, never one quarter's total with another's, so a quarter that is still being filed cannot look like selling. Fund and manager names appear as reported in the filings.

Yields and spreads

Yield to maturity is solved from the price (plus accrued interest; fund values exclude it), coupon and maturity on a regular coupon schedule: semiannual for US-dollar bonds, annual for euro and other annual-coupon currencies. It is not computed for floating-rate, defaulted, payment-in-kind or convertible bonds, or bonds with no maturity. Callable bonds are measured to maturity, not to their call date.

US-dollar spreads are measured against the Treasury curve interpolated at each bond's remaining life. The curve is FRED's monthly average of daily constant-maturity yields, so a month-end mark is compared with that month's average; this adds roughly 10–20 bp of noise to month-to-month changes. Euro spreads are measured against a Nelson-Siegel curve fitted each month to German federal bonds held by funds (inflation-linked bonds and outliers excluded).

"Company versus its government": a company bond's yield minus a Nelson-Siegel curve fitted to its own government's bonds, same currency and month, at the bond's maturity. Only plain corporate bonds count (covered and secured bonds, supranationals and state-owned issuers are kept apart), each with at least $25M held and three funds pricing it; gaps beyond ±100 bp are held back for review.

Priced-in default odds (bond market)

Annual hazard ≈ spread ÷ (1 − recovery), with 40% recovery; five-year odds = 1 − e^{−5 × hazard}. A spread also pays investors for liquidity and for bearing risk, so these numbers are higher than the share of borrowers that actually default. They are labelled "priced-in" for that reason. Governments borrowing in their own currency get no odds figure: they can print it, and the risk shows up as inflation and currency moves instead.

Distance-to-default (stock market)

For every listed non-financial company with a debt figure: a Merton/KMV model solves for the value and volatility of the company's assets from its stock market value, one-year equity volatility, a debt barrier of short-term debt plus half of long-term debt, and the one-year Treasury rate. Distance-to-default is the number of standard deviations the assets sit above the barrier over one year. Pages show it as a percentile among all companies ("safer than N% of companies"). The model also gives a probability, N(−distance-to-default), but that figure is risk-neutral and overstates real-world odds, so it appears only here. Banks, insurers, BDCs and REITs are excluded: leverage is their business and the model misreads it.

Validation before publication. Over 4,871 companies and 208,896 company-months (2020-12-31 to 2025-09-30), distance-to-default ranked companies that went on to a bankruptcy filing within 12 months (156 companies; 8-K item 1.03 or a Q-suffixed ticker) ahead of the rest with an area under the curve of 0.887 (month-by-month average; 0.886 on the common sample). Shuffled labels gave 0.50 on average (p = 0.002). The setting published (252-day volatility, half of long-term debt in the barrier) sits in a flat region of the tested grid, not at its single best point.

The shortest-distance decile held 67% of the later bankruptcy filings.

Altman Z (Z for manufacturers, Z″ for others) is shown as an accounting cross-check from the latest balance sheet; it ranked later bankruptcy filings less well (AUC 0.752).

Sovereign default case studies

Every sovereign default since 2019 that our N-PORT history covers, with the median dollar-bond spread and priced-in five-year odds 12, 6 and 3 months before. Our fund-mark history starts November 2020, so the 2020 defaults (Argentina, Ecuador, Lebanon, Zambia) are not covered. Case studies, not proof.

Russia's spread six months before default was about 157 bp; it widened after the February 2022 invasion and sanctions, not on fiscal data.

Issuers, ETFs, 13F and officials

Bonds are linked to a listed company by the CUSIP issuer number, the issuer's legal entity identifier as funds report it for the stock, or a ticker funds put on the bond line; name matching to a parent is used last and marked low confidence. Pages show only high- and medium-confidence links. ETFs are identified from the SEC's fund ticker list matched to each fund's N-PORT series. 13F data come from the SEC 13F filings: options are reported at the value of the underlying shares, not the premium; changes compare only managers who filed in both quarters. Officials' trades come from House and Senate periodic transaction reports and OGE Form 278-T; amounts are the disclosed ranges.

What these pages do not do

No ratings, no forecasts and no recommendations. Fund marks lag the market by about two months, so a bond's price here can be far from where it trades today. Nothing here is a live price except the exchange-listed notes.