Maple syrup pools in the June 2026 BTC drawdown

TL;DR — the decision rule

Same shock, opposite transmission. Through the BTC −21.5% / XRP −20.3% drawdown (May 22 → Jun 7-10), syrupUSDC’s realized APY stayed flat at ~4.6–4.8% — the fixed-rate loan book carries yield through a drawdown that compressed Morpho’s floating-rate equivalents by 60–80 bps (see ../2026-06-10-morpho-btc-stable-vaults-drawdown/). Share price rose monotonically every single day (1.165005 → 1.168078): zero credit loss, zero loans in Liquidated state, withdrawal queue never above ~$82k. What did move is the safety margin, not the return: pool-wide collateralization stands at 133.6% (API, 2026-06-10) — below the 136% Maple reported as its all-time low in the Oct-2025 stress — and the off-pool PYUSD reserve was drawn down ~$95M servicing the week. The rule:

In a sub-25% BTC drawdown, a fixed-rate custody-collateralized book converts the shock into cushion erosion instead of yield compression: APY and NAV hold, while collateralization absorbs the entire move. Hold through it, but monitor the pool collateralRatio (Maple API) as the primary stress dial — it is the analogue of Morpho’s utilization tell, and unlike Morpho there is no on-chain liquidation machine behind it, only Maple Direct’s 24-hour margin-call process at off-chain custodians.

Question

Our Morpho study showed BTC drawdowns reaching stablecoin lenders as utilization-driven yield compression with zero bad debt. Maple’s syrup pools are the other dominant venue for the same trade — stablecoins lent against BTC — but with inverted mechanics: fixed-rate loans, KYC’d institutional borrowers, collateral held at custodians (Anchorage, BitGo, Zodia) rather than on-chain, margin calls cured by humans within 24h rather than liquidated by bots within blocks, and a FIFO exit queue rather than atomic withdrawal (see ../../protocols/maple-syrupusdc.md). The endowment-relevant question: does the same shock produce losses, yield damage, or exit friction here — and which on-chain dials actually move when the collateral is off-chain?

Universe

Maple pools ≥$100M, both on Ethereum (Maple API + on-chain, 2026-06-10). Loan-book collateral split from the public openTermLoans book (per-loan principal × collateral asset):

Two-thirds to three-quarters of each book is BTC — comparable BTC intensity to the Morpho Base vaults (94–98%) once you net out the ~20–25% stable-collateralized legs. XRP adds a second volatile leg the Morpho vaults don’t have, and it fell as hard as BTC this window.

Methodology

Findings

The shock

BTC −21.5% peak-to-trough ($77,546 May 22 → $60,862 Jun 7); XRP −20.3% ($1.372 May 21 → $1.093 Jun 10, still at its low at window end). Both collateral legs fell together — the correlated-drawdown scenario the eval’s Risk #2 names.

APY: fixed-rate carried straight through

Realized APY (annualized day-over-day share-price growth, on-chain, 12-decimal precision): syrupUSDC held a 4.53–4.80% band with no trend — May 21 4.69%, Jun 10 4.70%, cumulative window growth 4.68% annualized. The single outlier (Jun 9, 5.10%) is an accounting true-up, not a yield event — traced on-chain: the loan manager (0x6aceb4…0fac) claimed five borrower payments that day, including a $17.6M loan repaid in full (tx 0x268cf902…, with $613 late interest) and a $1.27M interest payment on the book’s largest loan carrying $15,320 of late interest (tx 0xf5146890…). Scheduled interest accrues smoothly into totalAssets() via the issuance rate, but late interest is only recognized at claim time — the ~$16k of it that landed Jun 9 matches the excess share-price step (~1.3e-5/share ≈ $15k) and annualizes to the 5.10% print — a real one-day realized return (lenders did receive the penalty fee), but not a change in the pool’s forward earning rate, which is the distinction the chart annotation draws. Maple’s published APY, which is smoothed, correctly shows nothing there (DeFiLlama: 4.71% that day). Two real signals fall out of the trace: a borrower ran late on an interest payment during the drawdown week and cured with late interest — the only visible crack in an otherwise invisible off-chain process — and the $17.6M full repayment is Maple’s analogue of the borrower deleveraging we measured on Morpho. syrupUSDT likewise (3.89–4.42% band). DeFiLlama’s advertised APY tells the same story (4.80% → 4.62% → 4.72%). Over the identical window the Morpho Base vaults compressed ~80 bps. The mechanism: Maple’s loans are fixed-rate bilateral agreements — borrower deleveraging changes collateral, not the rate — so the drawdown cannot reach the lender’s yield until loans roll over. The flipside: when leverage demand collapses for longer, Maple re-originates at lower rates with a lag, where Morpho reprices instantly in both directions.

TVL and flows: drift, not run

• syrupUSDC −5.3% peak-to-trough (−$74.5M), modest and orderly — against the −41% Apr-2026 bridge-scare run, this barely registers. syrupUSDT −16.1% from its May 30 peak (−$68.7M) — the smaller pool saw proportionally heavier outflows.
• The queue never built: peak 70,000 shares ≈ $82k (Jun 5), zero by Jun 10. Every redemption cleared against incoming cash. No exit friction materialized at any point.
• But the buffer ran to $3.0M (0.23% of TVL) on Jun 8 — the thinnest single-day reading in the series — and the off-pool PYUSD reserve EOA was drawn down ~$95M (May 28 $352.5M → Jun 9 $257.0M, partially rebuilt +$15M on Jun 10) keeping it that way. Redemption capacity through the stress visibly leaned on the discretionary single-key reserves the eval flags as Risk #5, not on the pool’s own cash. (The companion reserve EOA 0x2570…3e08 holds only dust on mainnet — its ~$101M is DeBank multichain LP, untouched this window.)

Losses and margin calls: zero realized, invisible process

• convertToAssets() rose every single day of the window (1.165005 → 1.168078). No loss was socialized — same headline as Morpho’s zero bad debt.
• openTermLoans(state: Liquidated) returns zero loans, and no impaired/default state exists in the schema. No collateral liquidation happened through the drawdown.
• Maple’s framework (docs/FAQ, insights): when collateral falls to the Margin Call Level the borrower is automatically notified and has 24 hours to restore the Initial Collateral Level, else liquidation. With a 67–75% BTC book through a −21% move, margin calls were almost certainly triggered; whether they were issued and cured is not observable — collateralTxs is auth-gated, custody is off-chain, and no June-2026 disclosure had been published as of this writing (the precedent: Oct-2025 flash crash, 9 margin calls all cured within ~3h, per Maple’s CEO via news coverage and our eval’s Risk #2). The state-based evidence — no liquidations, no losses, ratio still >129% — is consistent with cures, but the process itself is a black box. Contrast Morpho, where we counted 4,800 collateral top-ups per day on-chain.

Collateralization: the cushion took the hit

The entire drawdown landed on the safety margin. Pool-wide collateralization (API, 2026-06-10): 133.6% syrupUSDC / 129.6% syrupUSDT ($1,739M collateral vs $1,304M principal). Two flags:

1. This is below the 136% that Maple reported as the all-time low during the Oct-2025 stress (and far below the 167% our eval recorded from Maple’s app on 2026-06-05). The two figures may not be methodologically identical (app metric vs API aggregate; the API number was not captured pre-drawdown), but on the API’s own terms the book is at its thinnest observed cushion. With ~22% of principal collateralized by stables (PYUSD/USDC, ratio ≈ par), the implied ratio on the volatile (BTC/XRP) book is roughly 140%: a further ~25–30% BTC/XRP leg down with no cures would push it toward ~100% — the underwater line — with only the 24h-margin-call process, not an on-chain liquidation engine, standing in between.
2. No collateral-ratio history is publicly available (no API timeseries, collateralTxs gated), so cushion erosion is only visible by polling — it is not currently among our monitoring checks for this protocol.

Decision framework

1. Fixed-rate books invert the drawdown transmission. Morpho: yield compresses, cushion defends itself via liquidations and borrower top-ups. Maple: yield holds, the cushion silently absorbs everything. Holding through a sub-25% drawdown was right in both venues — but for opposite reasons, and “APY unchanged” at Maple is not evidence of low stress. Check the cushion, not the yield.
2. The stress dial is collateralRatio, and it should be monitored. It is the only public, quantified read on how close the book is to the margin-call/liquidation cascade. This window took it from “comfortably above the historical low” to “below it” while every other depositor-visible metric looked serene. Proposed check: poll the API ratio; warn < 130%, page < 120% (vs 133.6% now, ~140% implied on the volatile book).
3. Exit capacity is discretionary, and the window proved it. The queue stayed at zero because ~$95M of single-key reserve PYUSD was deployed and the pool buffer was allowed to run to $3.0M (0.23%). That is the eval’s Risk #4/#5 operating as designed under mild flow stress (−5%) — it says nothing about a correlated allocator exit, and the buffer’s $3M day shows how little standing cash the pool itself carries.
4. Tail shape differs from Morpho. Morpho’s loss scenario is a gap move outrunning bots in minutes; Maple’s is a gap move outrunning a 24-hour human cure window at off-chain custodians, with XRP as a second correlated leg and rehypothecation risk on top (eval Risk #2). A −21% week never tested either. Do not read this study (or the Morpho one) as evidence about >30% gap scenarios.
5. Same systemic leg, both venues. syrupUSDC’s book (67.5% BTC) and the Morpho Base vaults (94–98% cbBTC) are the same macro trade — stablecoins against BTC leverage demand. Diversifying across them diversifies mechanism (liquidation engine vs margin calls, floating vs fixed), not exposure. Position sizing should treat them as one BTC-collateral-lending bucket.

Caveats

• Collateralization history is unobservable — the 133.6% is point-in-time; the pre-drawdown API baseline was never captured, and the eval’s 167% (Maple app, 2026-06-05) may be a different metric. Reconciling app-vs-API collateralization is the most useful follow-up; until then, treat the “below Oct-2025 low” comparison as indicative, not proven.
• Margin-call activity is inferred, not observed: collateralTxs is auth-gated, custody flows are off-chain, and Maple had published no June-2026 stress disclosure as of 2026-06-10. Zero liquidations + monotonic share price is the outcome, not the process.
• Loan-level LTVs and per-loan collateral values are not public — the BTC share is measured by loan principal, not collateral value; the implied ~140% volatile-book ratio assumes stable-collateral loans sit near par.
• Realized daily APY is lumpy by construction: Maple’s NAV recognizes some interest discretely (payment/claim events), so single-day readings can jump ±0.4pp on a flat book (Jun 9’s 5.10%). The band and the cumulative growth rate are the signal, not single days. The 6dp sharePrice columns in onchain-daily.csv additionally quantize (~±0.3pp); data/shareprice-hires.csv (12dp, pull-shareprice-hires.sh) is the precise series and is what the chart uses.
• DeFiLlama TVL for these pools is ~2.3× the on-chain figure (used for APY only); CoinGecko prices are third-party daily closes (accessed 2026-06-10).
• Off-pool reserve movements are read as balances only — the −$95M PYUSD drawdown’s destination (redemption funding vs repositioning) is inferred from timing, not traced transfer-by-transfer.