MoonUp

THE 65-HOUR PROBLEM: How Tokenized Equity Collateral Survives a Closed Market I. Two Clocks A tokenized stock runs on a blockchain clock, which never stops. The share behind it runs on an exchange clock. The @NYSE regular session covers 32.5 of a week’s 168 hours.  The weekend alone is 65 straight hours with no live market.  nyse.com/markets/hours-calen… A lending protocol that accepts stock tokens as collateral must still price them, margin them and liquidate them during those hours. This paper argues that closed market handling has moved from a theoretical risk to an operating requirement, because three major venues now accept stock tokens as collateral and at least two of them have already recorded weekend dislocations in related markets. II. Why the Weekend Breaks the Arbitrage On weekdays, issuance and redemption pull a token back toward its share price. That route is unavailable while the primary market is shut, so a Saturday discount cannot necessarily be bought and redeemed against the underlying shares.  The mint-and redeem route also belongs to a few parties. Coinbase says minting and redemption are handled by KYC onboarded institutional partners and Authorized Participants.  @LlamaRisk assessment adds that someone who acquires a @coinbase B20 token in the secondary market holds an unvested position and cannot redeem until an issuer controlled vesting process is complete.  coinbase.com/en-gb/blog/toke… nitter.cf/LlamaRisk/status/21046… The practical result is that the token’s weekend price is set by whoever trades on-chain. @CoinDesk describes the holder as having a claim but no market to convert it, so the premium to the last close is a risk the buyer carries until the open.  coindesk.com/markets/2026/09… III. Mechanism One: The Feed That Holds Friday’s Price @chainlink answer is a feed built for market hours. Its @RobinhoodApp feeds report token price as the underlying equity price times a multiplier read from the token contract.  The equity price comes from 24/5 feeds that combine regular, pre-market, post-market and overnight sessions.  @chainlink says the reported valuation may stay unchanged over the weekend and tells developers to check whether each feed fits their collateral market.  docs.chain.link/data-feeds/t… docs.robinhood.com/chain/ora… @aave Base hub shows what that design means in practice. Aave V4’s Equities Hub has accepted seven Coinbase stock tokens as collateral since September 25. @LlamaRisk specifies an operating window from Sunday 8 p.m. to Friday 8 p.m. Eastern. Outside it the feeds hold the last value, while deposits, borrowing and liquidation stay open.  aave.com/blog/coinbase-token… theblock.co/news/defi/2026-0… nitter.cf/LlamaRisk/article/2104… Take a hypothetical borrower who posts $10,000 of a token with a 65% collateral factor and borrows $6,000 USDC. In V4 the collateral factor sets both the borrowing limit and the liquidation threshold.  On Saturday the token falls 10% on a DEX. The feed still shows $10,000, so the protocol sees a healthy loan. At true value the collateral is $9,000, the threshold is $5,850, and the loan is already past it. Reporting notes that such a position may first become liquidatable when the feed resumes Sunday evening and incorporates the move in one update, and that interest keeps accruing in the meantime.  aave.com/blog/coinbase-token… IV. Mechanism Two: Pricing the Gap in Advance Aave’s response sits in the parameters. The seven tokens carry collateral factors from 65% to 79%. LlamaRisk’s stress method uses historical off-hours stock moves, allows a 0.5% gap between oracle and market, and accrues debt at 24% annually, the top of the USDC borrow-rate curve, across the longest closure. It assumes liquidation completes within five minutes of the next regular open.  The maximum liquidation bonus is 5.5%. LlamaRisk states that its history cannot describe a decline rarer than any in the sample. theblock.co/news/defi/2026-0… dailycryptobriefs.com/news/a… The exit also has limits. Sale depth on Base for each token was roughly $0.27 million to $1.08 million for a 2% price impact in a September 17 snapshot. The hub’s draw cap is $21 million.  A liquidator of a large position has to split the sale or find a party with redemption access. theblock.co/news/defi/2026-0… Other venues take different routes. Syndromics, a lending marketplace on Robinhood Chain, lists a long tail tier with a 30% maximum LTV, a 45% liquidation LTV and a 15 point closed market haircut.  @Morpho listed markets for Robinhood’s NVDA and SPCX tokens use a 62.5% liquidation LTV, and the SPY market uses 77%. Different markets can use different oracle addresses for the same token.  Morpho’s general rules do not set weekend safeguards for each Robinhood stock token market.  crypto-economy.com/robinhood… nitter.cf/punkquidity/status/211… V. Mechanism Three: Pausing and Corporate Actions Some applications handle the clock by stopping activity. @paboxoxyz published explanation says new borrowing, liquidations and collateral withdrawals against outstanding debt pause while the relevant stock market is closed, while repayments and collateral additions stay open.  That protects borrowers from a thin weekend price, but it also delays liquidation while debt keeps accruing. Corporate actions need their own handling, because the token’s terms change. Robinhood’s docs say the oracle for an affected token is paused during processing, and integrators should treat a pause as price temporarily unavailable, not as zero.  Chainlink does not provide corporate action calendar data or automated pause triggers. Robinhood coordinates pause timing and multiplier updates.  In a 10-for-1 split example, the feed freezes at the last good value, the multiplier moves to 10, and the token price stays continuous at $200.  The pause flag is advisory and not enforced on-chain, so Robinhood tells integrators to keep their staleness check as the primary guard.  Robinhood’s docs also tell L2 integrators to check the sequencer uptime feed before trusting a price. docs.chain.link/data-feeds/t… docs.robinhood.com/chain/ora… VI. Three Cases Case One: the AMC token. CoinDesk reports that Robinhood’s AMC token went from $2.55 to as high as $23.16 on the night of Thursday, September 3, nine times the $2.54 NYSE close seven hours earlier. It was back at $3.26 within the hour.  A collateral market reading that token spot would have seen a nine-fold value for a short window. coindesk.com/markets/2026/09… Case Two: Black Saturday, October 17 to 18. These were perpetual futures, not tokenized shares. Reports say a large sell order on Binance at 21:21 UTC on Saturday set off liquidations of $51.6 million in contracts tied to SOXL, SanDisk and SK Hynix, spreading to @HyperliquidX . Thin liquidity worsened the drop with US markets closed, and prices rebounded quickly.  Semiconductors had logged eight straight losing sessions, their worst streak since 2008, before the drop.  The mechanism is the relevant part: forced selling into a book with no live reference price. coindesk.com/markets/2026/10… Case Three: the quiet weekend. Over a weekend in late August, Coinbase’s four stock tokens traded within 0.6% of held Friday values across four Aerodrome pools, and no live B20 reserve existed yet on Aave V3.  One calm weekend shows that a design works in calm. It says little about the weekend that follows bad news. VII. The Structural Question: Which Error Do You Accept? Every design picks a failure. A frozen Friday price can overvalue collateral after negative weekend news. A thinly traded token price can push a borrower through a threshold because of a temporary discount with no matching trade in the stock.  Haircuts, such as LTV cuts, move the cost to the borrower. Pauses move it to lenders, who wait for liquidation at the open. Using a live token price moves it to whoever is on the wrong side of a thin book. VIII. Where the Research Points This paper’s thesis is that closure-aware risk becomes a standard line item in every stock-token collateral listing, next to LTV and liquidation bonus. Markets will differ in how they split the gap between borrowers, lenders and liquidators, and the differences will be public in their parameters. The first realized bad debt event from a weekend gap will set the benchmark. Until then, the numbers above are modeled buffers, not tested ones.