apyUSD — Risk Report

Significant risk · 4.0/10

What this actually is

apyUSD is the yield-bearing wrapper around apxUSD — deposit apxUSD, receive apyUSD shares, and the share value grows as Apyx’s STRC-backed collateral pays dividends. Ongoing yield is ≈13% APY, within STRC’s 11-15% indicated-rate range. Where apxUSD holders forgo yield for stablecoin functionality, apyUSD holders accept an exit window — a 3-to-20-day unlock with a declining fee — in exchange for the dividend pass-through, or take a DEX two-hop for retail-scale exits in minutes.

⚠️ The vault’s own mechanics are sound; the question is what sits underneath it. ⚠️ apxUSD has held at or above par continuously since 2026-09-17, 393 consecutive hourly readings — ⚠️ and not since 2026-09-11, which is what a headline reading suggests: the ratio crossed par 21 times between 09-04 and 09-17, with 49% of readings in that window still below 100%. It is above 100% on the netted collateral ratio, on the gross basis, and on an independent conservative re-mark, with an attested surplus of 1.56%. ⚠️ That is a recovery and now also a thin buffer, on a book that was in deficit through the summer. ✅ What it removes is the realised undercollateralisation; it does not remove the backing concern, which rests on concentration and on what the reserve is made of rather than on the ratio alone. apyUSD trades at a discount to its NAV. Everything that makes this report a 3.0 comes from one layer down.

Yield	Exit methods	Effective time-to-cash	Age	Chains
≈13% APY ongoing	DEX two-hop (retail) or 3-to-20-day unlock window (institutional)	Minutes (sub-$1M via DEX) or 3–20 days (canonical)	≈3 months	Ethereum, Base

1 · Stability — 5.5

Reference: NAV, denominated in apxUSD. apyUSD has no $1 peg — it is a NAV-accruing vault share, and the NAV is the apxUSD-per-share ratio, growing as the STRC backing pays dividends.

⚠️ Which is why a healthy NAV here is not a healthy position. ⚠️ The NAV is measured in apxUSD rather than in dollars, so the share can accrue perfectly while the unit it accrues in moves independently. Through the summer that unit was below par and the effect was real; apxUSD has been above par continuously since 2026-09-17, now attesting a 1.56% surplus. ⚠️ But the collateral ratio and the market price are different quantities and only one of them recovered: apxUSD still trades about 1.10% below $1, and has never traded at or above $1.00 in 722 readings. The unit this share accrues in is sound on the balance sheet and discounted in the market, which is what an asset with no atomic redemption looks like.

⚠️⚠️ And the axis is cut rather than raised, with every figure behind the old number having improved. The discount narrowed from about −9% to −1.34%, NAV rose 1.376 → 1.436830 and is strictly monotonic across 722 readings, and the underlying ratio went from roughly 95% to 101.56%. ✅ The cut is because the smoothness was never evidence. apyUSD’s NAV is total_assets / shares, so zero drawdown across 722 readings is a property of the formula as much as of the assets — a share count divided into the assets it represents cannot fall out of step with itself. ⚠️ And the mark behind those assets is thin three measured ways: 52.18% of the STRC bucket ($80,169,860) is marked at an undisclosed issuer price; the verifiable 47.82% is marked off a venue holding about $371K against a $70.8M position; and this NAV has already been inflated once from this exact source — the June post-mortem records an inflated dashboard NAV from a STRCx pricing bug. ⚠️ The only independent price signal has never agreed with it: 0 of 722 readings at or above NAV on the DEX measure, 0 of 702 on CoinGecko. The exposure is to whatever apxUSD is worth, in either direction — which is why the concentration below matters more than the ratio on any one day. apyUSD has traded at a meaningful discount to NAV — mid-single-digit to about −9% during stress. Treat the specific discount as a moving figure and read it live on the dashboard.

⚠️ A units trap worth naming, because the obvious calculation gives a badly wrong answer. Published supply is denominated in shares (131,316,281); published backing is in dollars ($185,967,679). Dividing one by the other returns 141.6%, which reads as a $54.7M surplus. That surplus does not exist. The correct comparison is backing against supply valued at NAV — $185,967,679 against $185,967,679 — a vault-share identity that is true whatever apxUSD is worth, and says nothing about solvency. If you are checking this vault’s numbers yourself, confirm the units before dividing.

On the headline yield: NAV jumped ≈33% in week 1 (Feb 20-27, 2026) from a one-time launch-seed event — donation-pattern apxUSD inflows from a small set of addresses. Since week 2 it has grown smoothly at roughly 13% APY. ⚠️ A new buyer earns the ongoing rate and does not capture the launch jump, and at the current collateral ratio that 13% does not compensate for the backing and exit risk underneath it.

2 · Backing — 3.0

⚠️ The vault is 100% collateralized by construction and that fact is meaningless. It is denominated in apxUSD, so shares always equal the assets they represent. The collateral ratio that matters to an apyUSD holder is apxUSD’s, and it ran below par from June until the autumn. It has held at or above par continuously since 2026-09-17, now attesting a 1.56% surplus — ⚠️ while apxUSD itself still trades about 1.10% below $1.

Two ratios circulate and they are the same book measured two ways — both the issuer’s:

basis	ratio	what it excludes
netted — the headline attestation	101.5602% (2026-10-04)	protocol-owned liquidity and minted-but-unsold inventory, removed from both sides
netted, earlier reading	98.0177% (2026-08-23)	same basis — shown so the move is dated on both ends

⚠️ This is not a disagreement and not a gap between the issuer and us. Both are published by Accountable, and the netted figure reconciles exactly: (reserves − POL − inventory) / (supply − POL − inventory) reproduces the attested headline from the raw bucket split, to within 2.6e-05 pp across 723 of 723 readings. ⚠️ That tests the arithmetic, not the inputs. It is an internal-consistency check on a single feed, not a second source — which is why the trust banner above still stands.

Our independent lower bound is 98.5289%, computed gross after stripping the roughly $190,970 premium at which on-chain STRCx trades over the underlying STRC NAV. Against the issuer’s gross figure that is 0.06pp stricter — what a conservative bound should be, and exactly the size of the premium removed.

The shortfall is one number, not two: $4,399,857.64 against supply of $312,073,514.82, read 2026-08-23. That is Accountable’s own surplus_usd, and it is simply supply minus reserves of $307,673,657.18.

⚠️ The composition is worse than the ratio suggests. Net of the Inventory line — minted-but-unsold apxUSD, an asset offset by a burnable liability rather than real backing — the reserve is roughly three-quarters STRC family (≈74%), more concentrated than a gross reading implies, plus ≈13% cash and ≈13% reflexive POL deployed against Apyx’s own assets and capped at 15% of reserves. STRC currently sits at about 60% of reserves, above the 55% single-issuer threshold this coverage flags elsewhere.

Backing is 2.0, inherited from apxUSD and the lowest axis here. ⚠️ The portion we can check independently marks better than the attestation; the portion we cannot check is the larger one.

3 · Liquidity & Exit — 3.5

Two exit paths, and which binds depends on size. The axis takes the worse one — here they land together at 3.5, and the binding constraint is that both terminate in an asset that trades below $1, which is a market fact and no longer a backing one.

Retail-scale, sub-$1M — DEX two-hop, minutes to cash. Sell apyUSD → apxUSD on Curve, then apxUSD → USDC. ⚠️ Both legs price the apxUSD discount in, so the cash received reflects the below-par value, not NAV. Trading on the apyUSD/apxUSD pool is sporadic and market-maker driven, and Apyx pulls its own depth off-hours by design. Backup venues exist on PancakeSwap V3 and smaller Uniswap V4 pools. Depth on the Jupiter-routed STRCx venue is thin at roughly $180K, which bears directly on the arithmetic one layer down.

Canonical — a 3-to-20-day window with a declining fee. Burn apyUSD, enter the UnlockToken window, and exit faster by paying more: the fee declines linearly from ≈3.5% at about 3 days to ≈0.1% at the full ≈20 days. It is not a flat 20-day cooldown. The max window was shortened from 30 to 20 days by the Apyx admin on 2026-04-15, verified on-chain, and remains admin-mutable subject to a 72-hour visibility window.

⚠️ The window is both an exit cost and the protocol’s anti-bank-run feature — Apyx’s June post-mortem credits it with preventing a run by disincentivizing the simultaneous exits that would have forced more STRC selling. For institutional sizing it binds, and it exposes the holder to several more days of collateral drift before the terminal apxUSD is even received.

Because the canonical path terminates in apxUSD, this asset inherits the underlying’s redemption model. Apyx’s “Apyx 2.0” plan to move apxUSD mint/redeem to a single Redemption Value floor would improve the terminal asset — but it is blog-only, not in docs or on-chain, and it changes nothing about this vault’s own window.

4 · Dependencies — 3.0

⚠️ This is a chain, not a diversified book, and every link is the same story.

apyUSD  ->  apxUSD  ->  STRC  ->  MSTR  ->  BTC
 vault      the unit    ~74% of    the      the
 share      it is       reserve    issuer   collateral
            denominated
            in

apxUSD — 100% of what this share is worth. Its overall score is what this axis tracks.
STRC family — roughly 74% of the reserve net of inventory, and about 60% on the issuer’s own live figure, above the 55% single-issuer threshold flagged elsewhere in this coverage.
MSTR, then BTC — STRC is a preferred instrument junior to MSTR’s convertible debt. A severe BTC drawdown compresses MSTR equity, which compresses the dividend stream this vault exists to pass through.

Strategy’s discretionary soft floor remains active, but finite. The October 12 disclosure records $192.4M for 1,932,145 STRC shares at about $99.58, leaving $354.8M authorized; STRC closed at $99.88. ⚠️ That is a bid under STRC, not a peg, redemption right, or contractual floor. It reinforces the near-par mark without restocking Apyx’s reserve; meanwhile USD Cash fell $833.4M → $640.5M, only $140.5M above the nearest trigger, and the restricted USD Reserve held at $4.88B. apyUSD inherits that two-sided update only through apxUSD; none of its wrapper mechanics changed. The issuer release is live, but EDGAR still ends at the October 5 8-K; no October 12 accession is cited.

⚠️ Holding apxUSD and apyUSD together does not diversify. They are two claims against the same Apyx and STRC backing.

5 · Contract & Admin — 4.5

⚠️ This axis is set equal to apxUSD, and the reason is that it is literally the same admin surface. The three Safes described below are Apyx’s, not the vault’s — the wrapper adds a vault, and the vault adds no admin exposure of its own. ✅ A wrapper cannot be safer on admin than the thing it wraps when both answer to one set of keys, so this sits at apxUSD’s number rather than above it. The vault’s own cleanliness is real and is described below; it earns no uplift here because there is nothing to uplift past.

The vault itself is clean. A source review against a Sourcify full match confirms no privileged share-mint backdoor — issuance follows the standard ERC-4626 deposit path, with apxUSD transferred in before shares mint. The inherited risk comes from what depositors bring in, not from the wrapper. Three audits back the protocol (Quantstamp, Zellic, and Certora with formal verification); there is no bug bounty. The vault has had one observable implementation upgrade, about a month after launch, and future upgrades carry a 3-day window for the guardian role to cancel.

⚠️ The admin finding is not in any single row. It is in the intersection. Measured on Ethereum, three separate Safes govern the Apyx surface — and all three owner sets are identical: six shared, zero exclusive to any of them.

Safe	Threshold	Function
`0xABdd8c8e…65e96`	4 of 6	Token governance (AccessManager admin), behind the 72-hour timelock
`0xf9862EfC…3cE2`	4 of 7	Cross-chain bridge governor, no delay
`0x37b0779a…a555`	3 of 6	STRCx collateral custody — 705,956 STRCx, read on-chain 2026-09-09, no delay

⚠️ So the 72-hour timelock is not a control over these people. It is a control over one of the three doors they hold. Four of the six move the token and wait three days. ⚠️ Any THREE of those same six move the STRCx collateral and wait for nothing — the custody Safe is 3-of-6 on an owner set identical to the token admin’s. A delay is only a delay if the people it binds have no faster door.

⚠️ The size of that custody door is stated in tokens rather than dollars, deliberately. The Safe holds 705,956 STRCx — 42.0% of STRCx supply — read on-chain 2026-09-09. A dollar mark on it depends on which STRC price you use and on the date you use it, and a USD figure for this position can silently pick up a separate off-chain raw-STRC balance. The token count is the durable number; a USD figure for this door should be quoted with its price basis and its date or not at all.

⚠️ The bridge leg was the same shape until 2026-08-31 and is no longer. That Safe moved from 3-of-6 to 4-of-7 at block 25,877,589 — one owner added and the threshold raised — so crossing chains now needs the same four signatures as moving the token, not three. ✅ On the bridge specifically the quorum gap is closed and what remains is a delay gap: the token side waits 72 hours, the bridge waits for nothing. The custody leg is where the quorum gap still lives, and it guards the larger number.

⚠️ Read the evidence boundary before weighting this. The owner-set identity and the thresholds are measured on-chain and are the hard fact. Which function sits at which address is the softer half — the bridge attribution is read from a 2026-08-22 walk rather than re-derived, and the custody attribution comes from this coverage’s own STRCx records. If a role label here is wrong the concentration finding is unaffected, because three identical owner sets is what drives it, and that is the measured part.

⚠️ This is the same shape as the frxUSD finding, and it generalises: per-contract rows can each be correct while the composite is the risk, and the composite exists only in the comparison. Three rows reading 4-of-6, 4-of-7 and 3-of-6 tell you nothing about how many distinct people stand behind them. That number appears only when you intersect the sets, and nothing in any row prompts you to.

Read on-chain 2026-09-09, the token-admin and custody Safes hold the identical six owners, and the bridge Safe holds those same six plus one more. So the union is seven and ⚠️ the intersection is still six — and those six reach all three doors, because four of them satisfy the token Safe, three of them satisfy custody, and four of them satisfy the bridge. The honest count of people standing between an attacker and all three layers is not nineteen. It is six. The seventh owner adds a signer to the bridge; he does not add a door that the six cannot open.

6 · Issuer — 5.0

Same protocol, same team, same admin as apxUSD, and the axis carries the same 5.0. ⚠️ That is an inherited figure, not an independent judgement: if apxUSD’s Issuer score moves, this must move with it. The condition is stated because a borrowed score with no stated condition does not age into vagueness — it ages into a confident wrong number that still reads like a considered one.

In its favour: DFDV (Nasdaq-listed) backing with tier-1 investors including ParaFi and Pantera; Wolf & Company AICPA-standards attestations published monthly; a continuous TEE-attested proof-of-solvency feed at accountable.apyx.fi; and Alpaca named as the brokerage. The issuer named on the Wolf attestation is Preference Foundation, with Director Carolyn Kelly signing — Apyx as a legal entity appears separate from DFDV, a standard offshore-RWA structure.

⚠️ Against, and this is what steps the axis down to 5.0: Wolf is mid-tier rather than Big-4, and the April 2026 engagement narrowed to securities only — cash dropped from scope and may return. The June depeg exposed manual off-chain plumbing too slow to defend the peg, secondary depth discretionarily pulled off-hours, and lagging communications.

What the same event validated, and why the overall score sits at apxUSD’s level rather than below it: the apyUSD/apxUSD Morpho market took zero liquidations — its oracle keys off the redemption rate rather than spot — the one-way yield ratchet held, and no Morpho market booked bad debt. Demonstrated vault-level resilience against inherited backing damage; the two roughly cancel. (Separately, some other Morpho markets ran a stale self-managed oracle that lagged as apxUSD left $1, and are migrating to Chainlink.)

Who should avoid

Anyone treating this as a stablecoin substitute. It is a yield-bearing vault share, currently trading at a discount to NAV (axis 1).
Anyone sizing above what secondary depth absorbs. Above ≈$1M, slippage on the two-hop may force the 3-to-20-day window — and the terminal asset still trades below $1 even though its collateral ratio no longer does (axis 3).
Anyone uncomfortable with the launch-NAV structure. Early holders captured a one-time ≈33% bump that new buyers do not (axis 1).
Anyone holding apxUSD already. The two do not diversify — same backing, same issuer (axis 4).

What to watch

apxUSD collateral ratio back to ≥100%, sustained. Recovery here tracks recovery there (axes 2 and 4).
apyUSD discount to NAV — narrowing is healing, widening is renewed stress (axis 1).
STRC price and the soft-floor bid. Both the dividend stream and the backing recover if STRC re-rates (axis 4).
The Wolf attestation’s cash scope, which narrowed to securities only in April (axis 6).
MSTR and BTC drawdowns — STRC dividends compress in stress (axis 4).

A note on the apxUSD companion

The apxUSD report covers the non-yield-bearing sibling — the asset this vault wraps. apxUSD exits faster at any size, via Curve or USDC settlement, but earns no yield; apyUSD adds the dividend pass-through and the unlock window. ⚠️ The two are claims against the same Apyx and STRC backing. Holding both does not diversify.

Revision history

2026-10-04 — the underlying’s backing recovered and Stability was cut anyway; Stability 6.5 → 5.5, Backing 2.0 → 3.0, Overall 3.0 → 4.0 (Liquidity & Exit 3.5, Dependencies 3.0, Contract & Admin 4.5, Issuer 5.0 held). apxUSD’s attested collateral ratio holds 101.5602% continuously since 2026-09-17, which carries Backing. ⚠️ Stability moves the other way, and not for deterioration: the discount narrowed to −1.34%, NAV rose to 1.436830 strictly monotonically, and the underlying ratio reached 101.56% — every figure behind the previous number improved. The cut is because NAV smoothness on a total_assets / shares definition is substantially definitional, the mark under it is 52.18% issuer-priced with the verifiable remainder struck against about $371K of venue liquidity, and no independent price measure has ever printed at or above NAV — 0 of 722 on the DEX, 0 of 702 on CoinGecko. Redemption folds into axis 3; no number moved, redemption 3.5 against liquidity 3.5.
2026-08-30 — correction carried forward from the body; no score change. An earlier pass on 2026-08-23 stated that our independent bound read higher than the issuer’s attested ratio, and concluded the portion we can check marks better than the attestation. ⚠️ That was an artifact of comparing our gross bound against the issuer’s netted ratio — a +0.51pp illusion, not a finding. Like for like, our bound is 0.06pp stricter, as designed. The same pass published a second, larger shortfall of about $6.2M, derived by multiplying the netted ratio by gross supply; that figure does not exist on any basis and is withdrawn. The single correct figure is $4,399,857.64. ⚠️ The terminal asset’s concentration is unchanged either way, so nothing re-rates on this — and the collateral ratio, which was below par when this was written, has since crossed back above it without the concentration moving.
2026-08-25 — signer concentration measured; scores held at 3.0. Three Safes govern the Apyx surface: token governance 0xABdd8c8e…65e96 (4 of 6, behind the 72h timelock), bridge governor 0xf9862EfC…3cE2 (3 of 6, no delay), and STRCx custody 0x37b0779a…a555 (3 of 6, no delay, about the STRCx custody Safe). ⚠️ All three owner sets are identical — six shared, none exclusive to any of them. Six people holding three hats, so any three of the six move tokens cross-chain or move the collateral with no delay, and the timelock governs one of the three paths rather than the people. Owner-set identity and thresholds are measured; the role attributions are read from records.
2026-08-23 — apxUSD collateral basis clarified; scores held (Overall 3.0 / Underlying 3.0). The attested collateral ratio of 98.017743% is a netted figure, excluding $50.20M of protocol-owned liquidity and $39.92M of minted-but-unsold inventory from both sides, and is published alongside gross totals. The shortfall on gross supply is $4,399,857.64. An independent lower bound reads 98.5289%, computed gross after stripping the STRCx premium over STRC NAV. The terminal asset’s recovery gate is par or better on the attested feed, sustained. ⚠️ That gate was not met at this date, and the early readings that looked like it were noise: the ratio crossed par 21 times between 09-04 and 09-17. The clean break is 2026-09-17 — see the 2026-10-04 entry. What can be said is that the condition has started being met rather than remaining unmet.
2026-06-29 — STRC soft-floor reframe. See the apxUSD report for the reserve and team-trust write-up this inherits.

Live backing dashboard

Live data Open in new tab ↗

External dashboard

Open Live backing dashboard ↗

Open live dashboard