Tokenized credit funds compared — Securitize STAC vs Janus Henderson / Anemoy JAAA vs Midas mWIN

Refresh note (2026-08-17). Supersedes the 2026-06-24 run: the “open float grew 17×” finding is corrected (that inflow was Ethena’s escrowed $200M — a second anchor, see Finding 3), LlamaRisk’s June evaluations of both funds are folded in, and Midas mWIN joins as the third fund.

TL;DR — the decision rule

This study now compares three tokenized credit funds. STAC and JAAA hold the same asset (senior AAA CLO tranches, floating SOFR + spread, 30%+ subordination, <0.1% historical default) and earn the same real carry at a rising NAV — between them the wrapper is the whole risk, and at equal size JAAA remains the safer wrapper: it caps theft behind a 4-of-9 Safe + 48h timelock and discloses its holdings, while STAC’s mint/burn/seize/pause/upgrade all sit on un-timelocked single Securitize keys. But the pair is no longer two positions: since June both funds are two-anchor books with the same two anchors — Grove (Sky’s USDS allocator) and Ethena (USDe backing) — STAC 29% Grove / 71% Ethena, JAAA 55% / 29% with only 16% genuinely open, and LlamaRisk caps the two against one shared allocation limit. Size the pair as one correlated position downstream of both USDS and USDe stress, against the anchors’ flows rather than headline AUM. Midas mWIN is the direct competitor to that pair — the same product shape (bankruptcy-remote wrapper, institutional custodian, daily NAV, daily redemption) around a Wellington-managed multi-sector credit book targeting about 5%. Head-to-head, mWIN wins exactly where the pair fails: it is anchored by neither Grove nor Ethena, and it is the only NAV of the three that honestly marks down. It loses nearly everywhere else — a single-key issuer plane on par with STAC’s, the tightest transfer gate of the three, and no auditor, fee schedule, or independent review, at twelve days old and $14.7M. Two blockers keep it the watchlist competitor rather than the third allocation: we have only evaluated lending against it (the Sentora Morpho vault), not holding it, and at its size a NOCA-scale ticket would itself become the fund’s anchor — the very concentration this study penalizes the pair for.

Question

We hold strategies for two tokenized AAA-CLO funds — Securitize STAC and Janus Henderson / Anemoy JAAA. A third tokenized credit fund, Midas mWIN, competes directly with them for the same structured-credit-RWA allocation; our only strategy touching it so far covers the venue lending against it, Morpho Sentora mWIN Main. The recurring questions when any comes up for sizing are (a) which wrapper concentrates more risk, (b) how durable is the capital beside us, and (c) what the carry actually is net of cost. This study answers all three with side-by-side data so no strategy has to repeat it, and so a sizing decision on one can reference the others.

Methodology

Findings

1. Same asset, same carry — different wrapper (and mWIN: a different asset)

STAC and JAAA are an identical economic engine; the differences are all in how the share is issued, governed, and disclosed. New since June: both funds now carry LlamaRisk coverage — the June run’s “no independent rating of STAC” is superseded. LlamaRisk also ran a pricing-consistency check: STAC marks its tranches 4–5bp above the public Janus Henderson AAA CLO ETF (never more than about 10bp), while the JAAA fund and that ETF price identically. mWIN breaks the “same asset” premise: CLOs are one sleeve of a broader Wellington credit book with an about-5% target, so its rows here compare the wrapper, not the credit — and it is the only one of the three with no auditor named, no independent review, and less than a month of history.

2. Governance & key-compromise — STAC and mWIN are single-key issuers; JAAA is not

The STAC/JAAA claims were re-confirmed in a Tenderly mainnet-fork simulation (2026-06-24): impersonating STAC’s real MASTER/ISSUER EOAs, pause(), burn(holder,…), and issueTokens(…) each succeeded with no signature, no timelock, no second signer. JAAA’s theft path provably requires the 4-of-9 Safe to clear the 48h Root delay. The mWIN column is carried from morpho-sentora-mwin-main (RoleGranted history + hasRole reads, 2026-08-14), not re-simulated here. LlamaRisk’s PPM read adds discretionary-power detail on STAC: the board may suspend NAV, subscriptions, or redemptions with no objective triggers, and may compulsorily redeem “for any reason or for no reason.” Net ranking on the funds tier: JAAA (4-of-9 + 48h) > mWIN (1 key instant, but a bankruptcy-remote compartment and 48h on code) ≈ STAC (1 key instant, everything) — with the caveat that mWIN’s single key controls the price every redemption and every downstream market relies on, a power neither BVI fund’s keys have in on-chain-enforceable form.

3. Capital retention & anchor concentration — the June “diversification” was a second anchor

How to read this. All three venues are dominated by holders that don’t trade day-to-day. The solid teal line is Grove’s share of JAAA’s Ethereum leg; the dashed teal line adds the second anchor (the Ethena/Solana escrow) once it appears in June. The coral line is mWIN’s equivalent concentration — the share of its supply locked in the Morpho market — which went 0% → 93% in the week after launch. It’s a concentration gauge, not a measure of withdrawable liquidity (exit is a gated off-chain queue either way — and a co-redeeming anchor worsens your liquidity by forcing CLO sales into that same queue).

What actually happened, corrected against the June run:

• STAC’s Ethereum leg is unchanged: Grove flat at 100,000 tokens, open float 295 tokens (≈$0.3M) — 99.7% Grove. But the fund tripled: Ethena’s Solana allocation (announced 2026-06-12) is now 244,961 tokens ≈ $252M, 71% of the $355M fund (rwa.xyz, 2026-08-17). STAC went from one anchor to two; open capital is still ≈0.1%.
• JAAA’s Grove did exit — that part of the June finding stands. Grove redeemed its Ethereum seed from $748M to $130M, including a single-day $318.6M redemption (42.8% of AUM) on 2026-03-11, settled at about 5bp slippage via a Bank of America block trade with no NAV impairment (LlamaRisk). That is the best crisis datapoint any fund in this study has.
• But the June inflow that replaced Grove was Ethena, not open demand. Of the 204M JAAA minted on Ethereum between May 25 and Jun 24, 192.8M went to the Solana-escrow EOA in one transaction on Jun 9 (Methodology). True open float went ≈$14M (Nov) → ≈$10M trough (Feb) → ≈$56M today — real growth of about 4×, not the 17× the June run reported.

mWIN’s column reads differently because its concentration is a different kind: not a redeeming anchor but a lending market — 93.1% of all mWIN sits in one Morpho Blue market as borrower collateral (ledger rebuilt from the token’s full Transfer history, 2026-08-17; up from 90.5% at the strategy’s 2026-08-14 read). It escapes the Grove/Ethena correlation entirely — but there is as yet no investor base to be retained: at $14.7M with effectively no open float, a direct holder of any NOCA-relevant size would instantly become the fund’s first real anchor, recreating on ourselves the concentration this study warns about in STAC and JAAA.

The decisive point is now two colors: Grove/Sky anchors both funds (≈$103M STAC + ≈$392M JAAA) and Ethena anchors both funds (≈$252M STAC + ≈$202M JAAA). Holding both funds stacks the same two decision-makers twice. The coral sliver on the right is mWIN drawn at the same scale — 1/48th of JAAA — which is why its sizing decision (Finding 8) cannot borrow this chart’s denominators.

4. NAV accrual — real carry; the CLO funds have never marked down, mWIN already has

Both fund NAVs rose monotonically across the entire window — including through Grove’s $618M exit and the March single-day $318.6M redemption — confirming the capital flight in Finding 3 was redemptions, not losses. JAAA’s pricePerShare is on-chain and daily; STAC’s NAV is off-chain (no on-chain getter; Chronicle’s Proof-of-Asset dashboard is live but we still have not read a feed on-chain), so only sparse rwa.xyz readings are observable. mWIN’s coral stub is its issuer feed divided by the $130,000 re-base value, so all three start at 1.00.

Rebased to a common start, JAAA has realized +2.99% (≈4.1% annualized) against STAC’s +2.80% (≈3.5% annualized). The gap has narrowed since June: over the last two months both annualize at about 4.5% (JAAA 1.038632 → 1.045520; STAC $1,021 → $1,028). Both are real AAA-CLO coupon (SOFR + spread), not emissions. (STAC’s line is a sparse off-chain series — treat its slope as an average, not a path.)

mWIN’s 19-day stub is steeper than either fund — and it is the only line that wiggles. The full feed since the 2026-07-29 per-token re-base (data/mwin-nav-feed.csv): $130,000 → $130,469.77 on 2026-08-17 — +0.36% in 19 days, about 7% annualized, ahead of Wellington’s about-5% target and of both CLO funds:

The path includes two down-prints (2026-08-11, −1.3bp; 2026-08-17, −2.6bp), and the pre-launch per-share era (feed rounds 1–10) contains two more — four down-days across the full feed history, against zero ever for STAC or JAAA. That is what an actively managed multi-sector book looks like beside par-floored AAA-CLO accrual: higher carry, two-way marks. Nineteen days is noise-level evidence for the carry; the down-prints are structural.

5. Cost — STAC is cheaper to hold; the exit toll needs a PPM re-read

Two corrections against the June run, both from LlamaRisk’s PPM read. Holding cost: STAC charges 0.30% management under a 0.40% total-expense cap; JAAA charges 0.40% management plus pass-through costs (brokerage, custody, admin, audit) — its all-in exceeds the 0.40% headline, so the June table’s “0.50% all-in” understated the comparison in STAC’s favor rather than overstating it. STAC remains cheaper to hold. Exit toll: the June run treated STAC’s 2.00% redemption fee as universal; LlamaRisk scopes it to the Liquidity Pool path, with the standard daily redemption priced off realized CLO liquidation (basis unstated in the PPM) and settling T+4 to T+5 — slower than the T+2/T+3 our strategy assessment carried. Until the PPM is re-read, treat STAC’s exit as: standard path slow and fee-basis-unclear, instant path 2% (plus a 24h lock-up).

mWIN’s cost column is mostly “not disclosed” — that is itself the finding: for a fund competing head-on with a 0.40%-capped and a 0.40%-plus-pass-through peer, neither the fund-level fees nor the instant-redemption terms are published, so the only observable economics are the net NAV slope (Finding 4). (The 15% performance fee in our related strategy belongs to the Sentora lending vault — a separate wrapper layer, irrelevant to holding mWIN directly.)

6. The shared anchors, traced — both funds sit downstream of USDS and USDe

The June run’s most important finding, now doubled:

Sky USDS → AllocatorVault (ilk = ALLOCATOR-BLOOM-A) → Grove ALMProxy 0x491e…a44e → 100,000 STAC (≈$103M) + 124.8M JAAA (≈$130M) on Ethereum, plus $250M JAAA on Avalanche. Grove has drawn ≈$2.76B USDS (re-verified 2026-08-17; was $2.65B in June).

Ethena USDe reserves → ≈$252M of STAC (Solana) and ≈$202M of JAAA (Solana, escrowed on Ethereum at 0x5d2c…d03b) — allocated June 2026 after LlamaRisk’s risk-committee approvals.

• Two on-chain entities anchor both funds. The correlation the June run flagged for Grove now applies twice over: a USDS stress event (savings-rate cut, depeg scare, sUSDS redemptions) forces Grove to redeem from both funds; a USDe stress event (redemption wave against backing assets) forces Ethena to do the same. Either drives CLO liquidation through the same gated, off-chain queue any NOCA position exits through.
• LlamaRisk independently reached the study’s conclusion: it treats STAC and JAAA exposure “as additive against a single shared allocation cap” (recommending about $310M combined on a conservative loss budget). That is third-party confirmation that the pair is one position, not two.
• The queue has one proven datapoint: the $318.6M single-day JAAA redemption cleared at about 5bp (Finding 3). It was cleared by a bank block trade in a calm market — do not extrapolate it to a correlated USDS + USDe stress, where both anchors hit the same queue at once.

7. Transparency & exit mechanics

8. mWIN synthesis — the direct competitor, not yet a direct option

Read as a holding, mWIN is the like-for-like competitor to STAC and JAAA: the same product shape (a tokenized credit fund — bankruptcy-remote wrapper, institutional custodian, daily NAV, daily redemption) sold to the same allocator, targeting about 5% against the pair’s realized 3.5–4.1%. Row-by-row it wins on the two things this study faults the pair for: it is anchored by neither Grove nor Ethena (Finding 3), and its NAV actually marks to market — four down-prints in the feed history against zero ever for the funds (Finding 4), honesty a par-floored accrual line cannot offer. It loses on nearly everything else: a single-key issuer plane on par with STAC’s, worst-in-class (Finding 2); the tightest transfer gate of the three (85 wallets, Finding 7); and no auditor named, no fee schedule, no holdings disclosure, no independent review (Findings 1, 5, 7) — twelve days old at 1/25th to 1/48th the size.

Two facts keep it out of the sizing decision today. First, it is unassessed as a holding: our only strategy touching mWIN (morpho-sentora-mwin-main) covers the Sentora vault that lends against it — a different exposure with a different risk book — so a direct position needs Midas greenlisting and a dedicated strategy that closes the fee, auditor, and redemption-depth unknowns. Second, scale inverts the anchor logic: at $14.7M with no open float, any NOCA-relevant ticket would make us the fund’s dominant holder — the exact concentration this study penalizes STAC and JAAA for, except we would be on the inside of it. mWIN is the name to re-compare once it has a real investor base, a disclosed fee schedule, or independent coverage — the watchlist competitor, not yet the third allocation.

Decision framework

• At equal size, prefer JAAA on wrapper safety. Same asset and carry, but JAAA caps theft behind a 4-of-9 Safe + 48h timelock, publishes on-chain NAV, and discloses holdings; STAC concentrates mint/burn/seize/pause/upgrade on un-timelocked single keys, with an affiliate NAV administrator and board discretion to suspend redemptions with no objective triggers (PPM via LlamaRisk). The credit is equivalent; the custody/governance/transparency gap decides.
• Treat STAC + JAAA as one position with two shared anchors. Grove (Sky/USDS, ≈$495M across both) and Ethena (USDe, ≈$454M across both) anchor both funds — ≈84–100% of each. A USDS or USDe stress event forces a shared co-redemption through the same gated queues. LlamaRisk sizes them against a single combined cap; so should we. Watch Grove’s flows, Ethena’s reserve disclosures, and USDS/USDe pegs as leading indicators.
• Price STAC’s exit properly: standard path is T+4/T+5 with the fee basis on the standard path unresolved; the instant path costs 2% — at a 3.5–4% carry that erases roughly half a year on a round-trip. Re-read the PPM before sizing anything above the strategy’s provisional $10M.
• Pre-enroll the access lists before depositing anywhere here — STAC’s and JAAA’s KYC allowlists cannot be joined during a crisis, and mWIN’s 85-wallet greenlist is the same constraint, tighter.
• Treat mWIN as the watchlist competitor, not yet an allocation. As a direct holding it is the only one of the three that escapes the Grove/Ethena correlation — but it is unassessed as a holding (the existing strategy covers only the Sentora lending vault, $2M as lender), fee-undisclosed, single-key, and so small that a NOCA-scale ticket would make us its anchor. Re-compare when it has a real investor base, disclosed fees, or independent coverage; until then do not let this study’s larger numbers anchor its sizing upward.

This framework stops applying when: STAC hardens its keys (timelock/multisig — re-rate immediately) or its Chronicle feed becomes on-chain-readable; either fund’s anchor mix shifts by more than 25% of AUM (Grove or Ethena entering, exiting, or being replaced); LlamaRisk’s shared cap or committee stance changes materially; JAAA’s open float rolls back toward the February trough (≈$10M); or mWIN gains independent coverage, loses its greenlist, or its feed regime changes. Any of these → re-run.

Caveats

• The Ethena attribution of JAAA’s Solana escrow is inferential: the balance matches Solana supply to the cent, the mint date matches the announcement, and the size matches the disclosed $200M — but the EOA is unlabelled and the custodian unknown. If it is some other party’s Solana-leg escrow, JAAA’s second anchor is misattributed (the concentration itself is unchanged).
• JAAA’s Avalanche Grove tranche ($250M) is third-party-corroborated (LlamaRisk, data 2026-05-09) but still not chain-verified by us. Base/Stellar/BNB/Monad legs (≈$56M combined) were not traced; some may themselves be escrow-backed rather than open.
• STAC’s NAV and its Solana/Ethena leg are off-chain / non-EVM — from rwa.xyz and press, not independently verified on-chain. STAC’s retention read rests on the Ethereum leg.
• Four PPM-level discrepancies are unresolved (our evals vs LlamaRisk’s PPM read): STAC settlement T+2/T+3 vs T+4/T+5; the scope of STAC’s 2% redemption fee; JAAA’s custodian (StoneX vs Pershing); the magnitude of JAAA’s pass-through costs. The LlamaRisk readings are used here as the more recent primary-document read; securitize-stac.md and janus-henderson-anemoy-jaaa.md need follow-up passes.
• The mWIN columns, series, and charts rest on twelve days of live history and a fee schedule that is not public (its yield is taken as net-of-fees at the NAV feed). Its comparison rows will move fast; re-check before relying on them.
• Smallest fact that would falsify the core call: STAC adding an on-chain timelock/multisig over its MASTER/ISSUER keys would collapse the governance gap behind “prefer JAAA.” Next-most-useful follow-up: re-read both PPMs to close the four discrepancies above, and verify the JAAA Avalanche tranche on its own chain.