Huma PST
Proof of Solvency dashboard captured 13 Nov 2026 · 15:23 UTC
Composite
Band
Gates
Disclosure Grade
Disclosure grade bands
Disclosure Grade Bands
Published disclosure grade
A - 2.70
Disclosure grade threshold
No gate binds. The composite is 2.9727, unchanged to four decimals since 2026-11-06: 0.27 above the A floor and 0.23 below AA. Two medium-impact omissions (items 5 and 25) cap the grade at AA, which does not bind. The letter rests on the methodology owner's reading of 6 November, applied unchanged: a matured note shown at full value while its contract is flagged is a gap, not a mismatch. Last week's note resolved as that reading expected; B-TFEUJPKS is treated the same way. With m = 1 the grade is BBB. The grade describes the blocked 10:31 snapshot, accurately stamped, and measures disclosure, not asset safety.
Per-dimension breakdown
Each score is a numeric evidence conclusion. Expand a dimension to inspect the item scores that support it.
- Composition by instrument4
- Evidence tier
- Reproduced / on-chain-verifiable
- Asset-category mix3
- Evidence tier
- Shown in detail or independently attested
- Counterparty identity per leg2.5
- Evidence tier
- Shown as a headline only
- Concentration2.5
- Evidence tier
- Shown as a headline only
- Underlying position volatility0
- Evidence tier
- Absent
How the score resolves
How individual item evidence adds up to the published disclosure grade.
ITEM EVIDENCE
25 × 0 - 4
Individual scores with evidence and a reading.
DIMENSION MEAN
9 scores
Item scores combine into each dimension's score.
COMPOSITE
2.97 / 4.00
Weighted average of the nine dimension scores
Negative-space map
What the page does not disclose, what each omission hides and how it affects a decision.
Matured notes at full value while their contracts are flagged in default
MaterialNot gated, m = 0, on the methodology owner's reading of 2026-11-06. B-TFEUJPKS matured on 13 November at 02:00 UTC and was flagged eleven seconds later; at 16:08 neither coupon nor principal was funded through its contract. The page counts USD 5,085,342 at full value without a status. Its replacement B-TFEUBN6A was minted to the Safe at 08:56:47 but is not shown. B-TFCYXTIU matured at 09:00 and was also flagged; its coupon was funded at 10:13 and the page shows par plus a separate USD 102,411 coupon entitlement, but no status or replacement. Last week's B-TFCYBH7W resolved as the adopted reading expected: coupon paid 80 hours late, replacement minted 13 hours after maturity, row removed. The principal reading remains unproven [UNVERIFIED]: no principal has been funded through any contract, and the flag has no published definition. With m = 1 this capture is BBB. A status on the row, showing the replacement at the roll, or a published definition of the flag with settlement status closes the question.
No mandate or allocation constraints
MediumNo eligible-asset list, venue list, cap or change process on the page, in a tooltip or in the payload. Over the week USD 24.7M left Arf, USD 25.3M was lent into two Morpho cbBTC/USDC markets, Jupiter rose by USD 11.9M, and after the snapshot USD 10.0M went into a new pool. None of it is bounded by anything a reader can see.
No volatility or drawdown disclosure
MediumThere is no volatility, drawdown or stress series anywhere, so a reader cannot tell how far any leg can move against the holder. Unchanged since 2026-07-19.
The TradeFlow book rolls and the page does not say so
MaterialEach note appears with a maturity date and a duration of "<=90 days". Twenty-seven notes, USD 150.0M face value, have matured since 30 April, and no principal has been funded through any contract. Twenty-two of the twenty-seven carry flagDefault. The Arf series carry footnotes that say "3-month rollover"; the TradeFlow row carries none. A reader cannot tell that the eNote position has behaved as a revolving facility rather than a ladder of self-liquidating notes, or what the flags mean. Arf publishes facility-level origination, payback, turnover and outstanding facilities, not performance.
No deviation bound on any leg's horizon
MaterialEvery leg carries a horizon: <=90 or <=14 days for the credit legs, "no fixed term" or "multiple terms" for the DeFi legs, with "withdrawals subject to available liquidity and access restrictions" on Morpho and "subject to liquidity and limits" on Jupiter. None states how far from NAV the leg could be realised. The TradeFlow horizon is a roll date in every observed instance [UNVERIFIED as to cash settlement off-chain].
No obligor look-through on 79.19% of reserves
MaterialArf (USD 195.87M) and TradeFlow (USD 111.27M) are named to series, tranche and issuing segregated portfolio. The borrowers are not named, counted or characterised. Five notes, USD 25.20M (6.50%), read "Issuer: Not provided"; debtAdditionalInfo() returns "Cayman Emerging Manager Platform SPC - EURO Trade Flow Fund SP" on-chain, a rendering gap. USD 25.31M of Morpho USDC supplies two cbBTC/USDC markets at 86% LLTV (USD 10.26M on Ethereum, USD 15.02M on Base), while the page reads only "Morpho, USDC". This is a counterparty the page does not name, not loan-type exposure under item 19's reading.
A reflexive strand not flagged
MaterialJudgment, not applied; open with the methodology owner. Jupiter Lend vault 45 borrows USD 21.40M of USDC against 22.65M PST: 5.27% of USDC lent across all 21 USDC-borrowing vaults and 4.57% of USDC supplied to the shared layer. Huma's pro-rata share at the snapshot is about USD 1.44M, 0.37% of reserves (USD 0.88M, 0.24%, on 2026-11-06), as the leg rose from USD 19.52M to USD 31.44M. The v3.1 cap for an unflagged leg allocated to a market collateralised by the issuer's own token is read here as a market built on that token. Whether pooled layers are in scope remains undecided; applying the cap would take item 3 to 1 and the grade to BBB.
Off-chain evidence is self-reported
MaterialThe Arf receivables rest on monthly Wadsack confirmations delivered as issuer-hosted DocSend links, unsigned. The latest is as of 2026-09-30; the October confirmation has not appeared and is not yet late.
The feed is blocked and the page does not say so
SoftThe page served the 10:31:07 reserve and supply snapshot at 15:24, 4 hours 53 minutes old, accurately stamped. Every poll from 15:27 to 16:15 returned the same blocked snapshot while Tori advanced every 15 minutes. The payload says coupon capture fails on B-TFEUQNV4 with "Unsupported eNote payment currency"; the Safe holds none of that EURC note. Parts of the payload kept refreshing. By about 16:20 neither published Jupiter wallet held jlUSDC; USD 10.02M had left Jupiter at 14:55 and USD 10.0M entered a new pool at 14:57 inside the same published wallet. The rest was not traced this pass. The accurate stamp means no gate was applied.
Two timestamps five hours apart
SoftThe Merkle root reads "Timestamp 2026.11.13 15:18:14" and "Signature was validated at 2026.11.13 15:24:03", beside reserve and supply figures from 10:31. The page does not say which figures the root covers.
Morpho rows do not name their markets or vaults
SoftNo Morpho row names its market or vault (USD 45.71M, 11.78%). The USDe row is collateral with no borrow, as on 2026-11-06; a borrow would make it a loop needing the v3.1 loop disclosure.
Unnamed valuation provider on 12.34% of reserves
SoftMorpho, Pendle and Merkl rows read "Provider-reported valuation", with position data four to thirteen hours old at the snapshot (6.02% of reserves on 2026-11-06). The Morpho values reproduce within 0.1%. Two immaterial Kamino rows (USD 1,526) read "Method: Not recorded".
Seventeen eNote marks were about 34 hours old at the snapshot
SoftSeventeen of 22 eNote marks were set on 12 November at 00:33, about 34 hours before the snapshot. Each states its own time.
The signing-policy statement does not describe the upgrade authority
SoftThe page states a 3-of-4 multisig over "most treasury wallets, program upgrade authority and pool ownership". The PST program upgrades through a different authority, unchanged.
Headline supply value sits above its two series
SoftThe headline supply value is USD 16,627 above the sum of its two series. One exchange rate is applied to both tokens, the conservative direction, as before.
The ratio is nominal
SoftThe ratio is nominal, and the page says so: "A haircut-adjusted ratio is not used for this dashboard." No stressed ratio is shown.
Quick wins
Eight quick wins, by impact per unit of effort.
Publish the mandate: eligible assets and venues, per-venue caps, and what change needs a signature
Publish a default and delinquency series, and the settlement status of every matured note (repaid in cash, rolled, extended)
Publish a volatility or drawdown series per leg
Publish a stressed, haircut-adjusted ratio beside the nominal one
Obtain signed or direct-from-source balance confirmations
Publish obligor look-through for Arf and TradeFlow: count, top-1 and top-10 share, sector and geography
State the expected deviation from NAV at each leg's stated horizon
State the upgrade quorum correctly and have a third party confirm the signing policy and halt arrangement
A PTR grade rates disclosure, not safety. It measures how much of what a risk professional needs is actually on the page: not whether the assets are safe, and not whether the displayed numbers are true. A dashboard can be fully backed and score poorly because it proves none of it; it can score well and still hold assets that later fail. This is not a credit rating.