Independent private credit valuation benchmark
Lenders moved73% further apartin six months.
On the same loans, not new ones — 583 of them, held without interruption by three or more managers. The median gap between the highest and lowest value reported for one private loan went from 51 to 96.8 basis points between Q3 25 and Q1 26 — a 90% widening, with 67.6% of those loans widening individually. Over the same stretch, dispersion in what the lenders originally paid fell 6%.
And on $68bn of private credit — 41% of everything we track — two lenders holding the same loan are more than two full points apart on what it is worth.
1,036 loans · 90 managers · every number defensible line by line
1,036 loans this quarter · 31 March 2026 · move the cursor to read any point
Three facts are enough
No jargon required. The disagreement is a structural feature of how private credit is reported.
- 1
One private loan
A company borrows from several funds at once. The loan is not traded, so no market sets its price.
- 2
Several lenders
Every listed fund must publish, each quarter and loan by loan, what it believes its share is worth. Those filings are public.
- 3
Different answers
Same loan, same day, different numbers. Nobody puts those filings side by side. That is what we do.
The blue band is every value the other 5 lenders put on this loan, from lowest to highest — across the quarters shown it never spans more than 0.9 points. Antares left it and is now 901 basis points below the nearest of them. Same loan, same day, same filings. 1 quarter is left out: fewer than half the lenders filed on this loan then.
Both ends carry the same manager, Antares — two of its funds, the same loan, the same reporting date, 1178 basis points apart. The entire gap on this loan sits inside one house.
17 filings behind this loan · every one of them on EDGAR
what we threw away
We dropped 5,320 positions before publishing anything.
That is 1.7% of everything comparable this quarter. Each one looked close enough to something else to be counted as the same loan — and wasn’t. A gap between two different loans is not a finding, it is a fabrication, so the pair gets broken and both sides disappear from every figure on this page.
See the one we refusedRYAN LLC
Two unrelated managers. Same borrower, same +350 bps spread to the basis point, same first lien. On the matching key, one loan.
The maturities are 46 months apart. Two lenders in the same instrument do not disagree by 4 years on when it comes due.
It happens on almost every loan
Ten loans from this quarter. Each line is one loan; the two dots are the lowest and highest value its lenders reported for it, on the same day. Borrower names are part of the subscription.
the gap between lenders on the same loan · median across every loan we track
Tap or hover any quarter to read it. The thin line under each bar is how many loans it rests on — the earliest reading carries 39% of today’s panel, so the early levels are not comparable. The long fall across this chart is coverage widening, not lenders converging. The move that is real is the last one: 47.9 bps in Q3 25 to 82.9 today, on a panel that barely grew.
They report income they never received
A fund books payment-in-kind interest as income and pays its dividend in cash. Subtract one from the other and the coverage ratio the market debates changes side.
the number nobody computes
On the reported measure only 53% look under-covered. The difference is payment-in-kind — interest a fund books as income and never receives. It lifts the coverage ratio without a dollar entering the account.
- Median, as reported
- 99.4%
- Median, cash only
- 90.6%
- Distributions measured
- $4.6bn
- Vehicles
- 94 of 206
Both lines are medians across the same vehicles, quarter by quarter. The gap between them — the part of reported income that never arrived in cash — has narrowed since Q3 ’22, from 15.8 to 8.8 points.
What moved is the level. Both measures fell — even the as-reported median now sits below 100% — which is why the share of vehicles that pay out more than they collect went from 71% to 81%.
145 of 206 vehicles tag all three amounts
Below 100% does not mean a dividend is unfunded. A fund may lawfully distribute out of realised gains or return capital. What it means is narrower and harder to argue with: the income actually collected in cash does not cover the distribution. Both figures are the fund’s own, filed with the SEC.
Who uses it
Four questions, one dataset. Every answer traces back to a filing.
“Our clients hold a perpetual fund priced at NAV. What is in the diligence file that says the NAV is right?”
“We are pricing an LP stake. The manager’s NAV is our denominator. Justify the discount with evidence.”
“Which book is marked richest against its co-lenders, and whose writedown lands first?”
“Is this manager’s valuation discipline defensible before we put clients into the next fund?”
What the index does not publish
The aggregate figure is published openly. What sits underneath it is the product.
- —The quarterly figure, citable
- —The full history since Q2 2023
- —Methodology and limitations
- —Chart pack for press use
- —Every loan, named — 1,036 this quarter
- —Each manager's signed deviation, by name
- —How fast each manager moves its marks
- —Point-in-time archive and silent revisions
- —API with provenance on every response
institutional access · priced per desk
Request accessMethod, and what it cannot do
Public, complete, versioned. The limitations are published with the same weight as the findings.
Source
XBRL Schedules of Investments filed with the SEC, plus the DERA business development company dataset — 170 funds across 12 quarters.
203,107matched declarationsMatching
Two positions are the same tranche when borrower, spread to the basis point and seniority agree. Maturity is a rejection test where filers publish it.
3fields must agreeIndependent check
The reported all-in rate is held out of the matching key, so it acts as a judge. Groups whose rates diverge by more than 50 bps, PIK netted out, are dropped. Across the history the flagged share runs from 12.1% on loans reported by two managers to 15.4% on those reported by four or more — the judge tests the widest rate gap in the group, so more holders means more chances to diverge.
11.1%dropped this quarterReconciliation
Each schedule must tie back to the balance sheet within 0.5%. Those that do cover $350.3bn of $531.9bn.
87/168funds tie to the balance sheet
Not a rating, not a forecast, not an opinion on any manager. It is the distance between values that other people published.
Quarterly and US only. The published index and every manager figure stay confined to loans reported by at least three managers, since a co-holder median needs a third opinion to mean anything. Loan-level gaps go one step further, down to loans reported by exactly two — a plain subtraction between two filed numbers, published only where the rate check corroborates the match. The SEC required BDCs to tag their Schedules of Investments from 1 August 2022 for funds eligible to file a short-form registration statement, and from 1 February 2023 for the others; the published history starts once the whole panel is tagged, which is our choice, not a regulatory date.
Press room
The figure, the method and the sources. No sign-up, no form.
Between Q3 25 and Q1 26, the median gap between the highest and lowest value reported for the same private loan widened from 47.9 to 82.9 basis points — a 73% increase on a panel that grew 4%. On $68bn of private credit, 41% of the total tracked, two lenders holding the same loan are more than two full points apart. Source: Kanonstone, from 145 SEC filings as of 31 March 2026.
On AmeriLife Holdings LLC, a $1.4bn private loan held by 17 funds across 6 managers, the gap between the highest and lowest reported value was 1178 basis points as of 31 March 2026 — 87.92 cents against 99.71. A year earlier the same loan showed 54 basis points. 1 of the 6 managers sits within five points of the bottom of that range; the other 5 mark it at least five points higher. Every figure is drawn from the funds' own SEC filings. Source: Kanonstone.
Key figures
- As of
- 31 March 2026
- Loans measured
- 1,036
- Vehicles
- 145 · 90 managers
- Assets covered
- $163.9bn
- Source filings
- 145
- Reconciled
- 87 / 168
Contact
Method questions, underlying data, verification of any figure.
Reply within one business day.
Verify anything here
Every figure resolves to a filing on EDGAR — accession number, page and tagged fact. No value is modelled, smoothed or estimated; where a filer omits seniority, the matching infers it from the same borrower and spread, and says so.
sec.gov/edgarWhat people ask before they trust it
Where the numbers come from, how the matching is tested, and who pays for the work. Every figure below is read from the same files as the sections above.
- 01What does Kanonstone measure?
- The gap between the highest and lowest fair value that different lenders report for the same private loan on the same date. Across 1,036 loans held by 90 managers, the median gap is 83 basis points, and 41% of the assets tracked sit in loans 200 basis points apart or more. It is a measurement of disagreement between published numbers, not a valuation of any loan.
- 02Where does the data come from?
- Public filings only. XBRL Schedules of Investments filed with the US Securities and Exchange Commission, plus the SEC DERA business development company dataset — 203,107 matched declarations across 170 funds and 12 quarters. No data is bought, and nothing that is bought would ever end up in what we publish. Every figure traces to the accession number of the filing it was read from.
- 03How do you know two filers are describing the same loan?
- Three fields must agree: the borrower, the spread to the basis point, and seniority. Maturity acts as a rejection test wherever filers publish it. The reported all-in rate is deliberately held out of the matching key so that it can serve as an independent judge — groups whose rates diverge by more than 50 basis points, payment-in-kind netted out, are dropped. That check removed 11.1% of candidate groups this quarter, and the survivors still show a 79 basis point median gap.
- 04How do you know the gap is real and not a flaw in your matching?
- Because the same matching rule produces two very different answers on two families of loans. Where filers themselves declare the price observable — level 2 in the fair value hierarchy — lenders land 27 basis points apart. Where they declare it comes from a model — level 3 — they land 104 apart, 3.8 times wider, on loans matched by identical code. The classification is the filer's own, never ours, and the ratio holds across 14 quarters while the quarterly panel grew 17-fold. A defect in the rule would hit both families equally.
- 05Could the gap be what lenders paid, not what they think it is worth?
- At a single date, partly — two lenders who paid different prices for the same loan can mark it differently and still agree on today's value. The movement rules that out. On the 583 loans held by the same lenders both on 2025-09-30 and now, the price gap widened 90% (51 to 96.8 basis points) while the spread in what those lenders originally paid for the same loans narrowed 6% over the same stretch. A shrinking cause cannot explain a widening effect. 67.6% of those loans widened individually, not just the median.
- 06Does the gap mean everyone disagrees, or just one reporter?
- Both happen, and the two are published side by side. Across the middle half of the holders, Q1 2026 loans span 35.4 basis points, against 82.9 between the extremes. On 17% of loans the gap comes down to a single reporter standing away from the others; on 31% the disagreement runs through the whole holder set. The reason this matters right now: between the last two quarters the extremes moved 65% and the middle half moved 82%, so the recent widening is not the work of a few outliers.
- 07Is this a rating, a forecast, or an opinion on a manager?
- None of the three. It is the distance between values that other people published. Kanonstone never produces a valuation of its own, because the day a model of ours touches a mark, the product becomes a liability rather than a reference. The published index and every manager figure stay confined to loans reported by at least three managers; loan-level gaps go down to two, where the comparison is a plain subtraction between two filed numbers.
- 08Who commissions the work, and who pays for it?
- Subscribers and the parties testing the marks. The quarterly corroboration file is commissioned by the auditor or the fund's board — never by the fund being tested. A manager who wants the same evidence for its own committee takes a seat on the platform and pulls it directly. What we will not do is sell a manager a report about itself: that is the incentive that discredited issuer-paid ratings.
- 09What are the known limits?
- Quarterly and United States only. The SEC required business development companies to tag their Schedules of Investments from 1 August 2022 for funds eligible to file a short-form registration statement, and from 1 February 2023 for the others; our published history starts once the whole panel is tagged, which is a methodological choice and not a regulatory date. Each schedule must tie back to the filer's own balance sheet within 0.5% to be publishable — 87 of 168 funds clear that gate, covering $350.3bn of $531.9bn. A manager may also hold a different tranche of the same loan, or value it on a different date within the quarter.
- 10What does it cost, and what is free?
- The market-wide index is free and stays free — it is meant to be quoted without a contract. Platform access starts at $15,000 a year per seat, or $50,000 for a five-seat desk. Engagements — corroboration files, pre-allocation reviews, secondary pricing reviews — are priced per decision.
83bps median gap
203,107matched declarations
3fields must agree
3.8×wider once a model decides
+90%while entry gaps narrowed
35.4bps across the middle half
0valuations of our own
neverthe fund being tested
87/168funds publishable
freethe market-wide index
Every number, back to its filing.
1,036 loans this quarter, 90 managers, as of 31 March 2026.