What happens when a stablecoin is unanchored for 30 seconds?

Summary

Most traders believe that unanchoring of stablecoins is a slow process, but this is not the case. In just 30 seconds, arbitrage robots, clearing cascades, and oracle delays collided with each other, turning a slight price slip into a systemic event.

Summary

Deanchoring of stablecoins that lasts less than 60 seconds could trigger hundreds of millions of dollars in liquidation in the DeFi space because lending agreements rely on price oracles that are updated at regular intervals rather than in real time, creating a window for outdated mortgage rates.

On Ethereum, the arbitrage robot can detect and use unanchoring in two to three blocks (about 24 to 36 seconds), buy discounted stablecoins in one place and redeem or sell them at a fair price at another. However, when unanchoring is triggered by solvency problems rather than liquidity imbalances, their speed advantage disappears.

Chainlink price increases for major stablecoins use a deviation threshold of 0.25% and a heartbeat interval of 1 hour, which means that the oracle will only update if the price has changed by at least 0.25% from the last reported value or 3600 seconds have passed, whichever comes first.

In March 2023, during the USDC unanchoring incident triggered by the collapse of Silicon Valley banks, approximately US$2.1 billion was liquidated in the DeFi field in the first four hours, of which the largest single liquidation amount on Aave v2 exceeded US$52 million. This is because borrowers using USDC as collateral have their positions below the maintenance threshold.

Curve Finance's 3pool (then the largest stablecoin liquidity pool on Ethereum) had its USDC balance soared from about 33% of the total pool to more than 83%. Traders sold USDC in exchange for DAI and USDT, a change in composition that exacerbated unanchoring by creating unilateral liquidity.

Common misunderstandings and realities of unanchoring

Popular explanations for unanchoring of stablecoins often involve a gradual loss of confidence: reserves are questioned, redemptions surge, and anchored exchange rates gradually disintegrate over hours or days. This version describes Terra's collapse, but cannot explain what happened when the USDC fell to $0.87 on Friday afternoon due to a bank failure, or the USDT briefly fell to $0.97 on the Curve during the liquidity crunch. These events last from seconds to minutes, and the damage they cause runs on completely different timescales that are far beyond the afterthought. The mechanism for short-term unanchoring is faster, more automated, and has more severe consequences per second than almost any other event in cryptocurrencies.

It is wrong to assume that unanchoring must last for minutes or hours to have an effect. Thirty seconds is enough for the automated system to declare the position insolvent, perform liquidation, sell the confiscated collateral at a discount, and then continue. Thirty seconds is enough for the liquidity pool to absorb a sell order large enough to shift its composition from balanced to severely one-sided. And 30 seconds is more than enough for the arbitrage robot to determine whether the unanchor is a buying opportunity or a real solvency event-a difference that determines whether the robot will stabilize prices or accelerate declines.

Anchoring mechanism under normal circumstances

stablecoins maintain their anchor with the U.S. dollar through a combination of primary market redemption and secondary market arbitrage. The primary market is the place where authorized participants (usually trading companies with a direct relationship with the issuer) can create or redeem stablecoin tokens for US$1 of underlying reserve assets. The secondary market is where everyone else trades, including centralized exchanges, decentralized exchanges, and automated market makers.

When the secondary market price is less than US$1, the arbitrageur buys a discounted stablecoin in the secondary market and then redeems it at US$1 through the primary market to earn the difference. When the price is above $1, they do the opposite: minting new tokens for $1 and selling them at a premium in the secondary market. This two-way arbitrage normally anchors the price at $1.

The system is effective because the primary market serves as the price floor and ceiling. As long as anyone can redeem $1 of reserves with 1 USDC, the token cannot be traded significantly below $1 for long periods of time, as doing so would create a guaranteed profit for anyone willing to execute the redemption. The key phrase is "long-term". The delay between detection and unanchoring and execution of redemption is where everything happens.

30-second anatomy of unanchoring

Typical unanchoring events on Ethereum unfold on a compressed timeline that most observers can only reconstruct after the fact.

Zero second: A large sell order impacts the stablecoin liquidity pool on a decentralized exchange. The order size is enough to change the pool composition and push the implicit price of the stablecoin below $1. On Curve Finance, which uses a special bond curve optimized for assets that should be traded at similar prices, a sell order of $10 million to $50 million can shift the implied price by 0.5% to 3%, depending on pool depth.

Seconds 1 to 6: The transaction is confirmed in the next Ethereum block. The new composition of the pool is now publicly visible. Every robot that monitors memory pools and block flows sees price deviations.

Seconds 7 to 12: Arbitrage robots operating across multiple venues detect price differences between decentralized exchanges and centralized exchanges where stablecoins are still trading at parity. The fastest robot submits a transaction in the next block, buying discounted stablecoins on DEX and selling on CEX.

Thirteenth to twenty-fourth seconds: Carry trade execution. If the initial sell order is a single event-large funds rebalance their portfolios or panic sellers liquidate their positions-the arbitrage flow absorbs the price effect and the anchored exchange rate recovers in two to three blocks. This is a benign scenario and accounts for the vast majority of stablecoin price deviations.

Seconds 25 to 30 and beyond: If selling pressure continues, the arbitrage flow cannot keep up. Robots are limited by their own capital, their willingness to hold inventory risks, and the speed at which they can move funds between different locations. When unanchoring continues to exceed arbitrage capacity, the market transitions from a liquidity event to a confidence event, and the dynamics undergo fundamental changes.

Oracle delays and clearing triggers

The most critical consequence of short-term unanchoring is not price fluctuations per se, but the interaction between that fluctuations and the oracle system used by the DeFi Lending Agreement to evaluate collateral.

Loan agreements such as Aave, Compound, and Maker do not use real-time market prices. They use oracle feeds (most commonly provided by Chainlink) that aggregate prices from multiple sources and update them on the chain based on specific rules. For major stablecoins, Chainlink's price estimates usually use a deviation threshold of 0.25% and a heartbeat interval of 3600 seconds. When the price changes by more than 0.25% from the last chain value, or one hour has passed since the last update, the price will be updated, subject to the first trigger condition.

This design is intentional. Updating on each block will result in high Gas costs and make the oracle vulnerable to short-term price spikes. But the design creates a window of vulnerability during unanchoring. If USDC is trading at $0.99 in the secondary market, but Oracle last reported $1.00 and the deviation threshold has not been reached, the agreement will still value USDC collateral at $1. Borrowers holding USDC as collateral will be given a grace period of several minutes until the oracle catches up.

When the oracle was finally updated, the impact was sudden. Every position that just above the clearing threshold at the old price may suddenly fall below the threshold at the new price. The agreement will not liquidate positions one by one in the order of risk. It opens all eligible positions to all liquidators simultaneously, creating a wave of clearing transactions that compete for block space and drive up Gas fees, thereby increasing the cost of executing carry trades, thereby reducing the arbitrage flow that would otherwise help stabilize prices.

This feedback loop-unanchoring triggers oracle updates, triggers liquidation, triggers more selling, triggers deeper unanchoring-is why short-term unanchoring can cause damage that is disproportionate to its duration. The USDC incident in March 2023 generated approximately $2.1 billion in liquidations in the DeFi space. The de-anchoring lasted for approximately 48 hours in total, but most of the liquidation occurred in concentrated bursts corresponding to the oracle update cycle.

Curve pools and unilateral liquidity

Curve Finance occupies a unique position in stablecoin infrastructure because its automated market makers are specifically designed to trade same-priced assets. Curve's stable exchange invariant (a mathematical formula that concentrates liquidity around the 1:1 price ratio) allows large transactions to be made at minimum slip points under normal circumstances. During unanchoring, this same design amplifies the problem.

When a trader sells an unanchored stablecoin into the Curve pool, the pool absorbs the sell-off by accumulating more unanchored assets and allocating more of the other assets in the pool. As the composition changes-for example, from 33/33/33 for the three-asset pool to 80/10/10-the implied exchange rate non-linearity for most assets deteriorates. For a pool that can handle US$50 million in swaps with a 0.1% slip under an equilibrium composition, when an asset accounts for 80% of the pool, the same swap may require a 5% slip.

This dynamics means that the Curve cell acts as both a stabilizer and an amplifier. In the early seconds of unanchoring, the pool absorbs the sell-off, and the stable exchange invariant brings prices closer to parity. As the composition becomes increasingly one-sided, the pool begins to amplify unanchoring by making it increasingly expensive for arbitrageurs to buy discounted assets. Liquidity providers (who deposit all three assets in an equilibrium allocation) find themselves holding mostly unanchored assets, a volatile loss that can become permanent if unanchored is not reversed.

When the arbitrage robot stops buying

The key turning point in any unanchored event is the moment the arbitrage robot stops providing a price floor. Robots buy unanchored stablecoins because they expect to be redeemed for $1 or sold at parity elsewhere. Their willingness to do so depends on two assessments: whether the issuer can actually cash in on the redemption and whether the capital needed to execute the arbitrage is worth the risk.

During the USDC's unanchoring period in March 2023, Circle had approximately US$3.3 billion in deposits in Silicon Valley banks, which accounted for approximately 8% of USDC's total reserves at the time. When SVB failed, the question was not whether Circle could eventually recover its funds, but whether Circle could process the redemption immediately. The arbitrage robot, which would normally buy USDC at $0.95 and redeem it at $1, stopped buying because the redemption mechanism was temporarily frozen over the weekend.

This creates a gap between the fundamental value of the stablecoin (depending on whether the FDIC leaves depositors intact) and the market price (reflecting liquidity that is immediately available for redemption). The gap persisted until Sunday night, when the Federal Reserve and FDIC announced that all SVB depositors would be fully compensated. The USDC price returned to $0.99 within minutes of the announcement.

The lesson is: Arbitrage can only provide a price floor when the redemption mechanism is operating normally. When the floor disappears, prices are determined entirely by supply and demand in the secondary market, which in a crisis is dominated by sellers.

What does the lending agreement see during the de-anchoring period

From the perspective of the lending agreement, the de-anchoring of stablecoins creates a series of specific risks. The risk parameters of the agreement are designed to deal with these risks, but are only effective within a certain range.

When the borrower mortgaged USDC to borrow ETH, the agreement maintains the loan-to-value ratio. If the USDC is valued at US$1 and the LTV cap is 80%, the borrower can mortgage US$100 of USDC to borrow US$80 equivalent ETH. If the USDC oracle price falls to $0.90 and the current value of the collateral is $90, the effective LTV rises to 88.9%, exceeding the liquidation threshold.

The agreement opens the position to a liquidator, who pays off part of the borrower's debt and obtains collateral at a discounted price, typically 5% to 10%. The liquidator profits from the discount, agrees to recover the borrowed funds, and the borrower loses some of the collateral. In theory, the mechanism would maintain agreement solvency even when the value of collateral declines.

In practice, this mechanism relies on the liquidator being willing and able to execute quickly enough. During the unanchored period, liquidators must buy unanchored stablecoins to pay off debts, which means they are absorbing the same assets that the owners are trying to sell. If liquidations exceed the market's ability to absorb sell-offs of unanchored assets, the agreement may accumulate bad debts, i.e., collateral values are lower than the debt value and there are no positions that the liquidator is willing to close.

Aave v2 accumulated approximately US$1.6 million in bad debts during the USDC unanchoring period, which is small compared to its total lockup volume, but demonstrates the existence of this failure pattern. A more severe or longer unanchoring period will generate correspondingly more bad debts.

What is not covered in this article

This article does not address the unanchoring of algorithmic stablecoins, which involves fundamentally different mechanisms. Terra's May 2022 collapse was caused by the failure of the algorithmic stabilization mechanism itself, rather than a temporary liquidity event or external reserve shock. The dynamics of algorithm unanchoring involve a death spiral between stablecoins and their paired governance tokens. This phenomenon does not apply to fiat currency guaranteed stablecoins such as USDC or USDT.

This article also does not cover the regulatory impact of unanchoring. Related bills and stablecoin legislation address reserve requirements and redemption rights, but the interaction between these requirements and real-time market mechanisms during the de-anchoring period is another topic.

Finally, this article does not cover the specific details of the risk parameters of each agreement. Each loan agreement sets its own oracle source, deviation threshold, liquidation bonus, and bad debt processing procedures. The general mechanisms described in this article are widely applicable, but the exact numbers and results vary from protocol to protocol.

Practical checklist

If you hold stablecoins or use them as collateral in DeFi, the following factors determine your exposure to short-term unanchoring events:

Check the source of the oracle in the lending agreements you use. Agreements that rely on a single oracle and have high deviation thresholds are more vulnerable to delayed clearing triggers. Agreements that use multiple oracle engines or more tightly update thresholds will reflect price changes more quickly, which may be protective (faster clearing prevents bad debts from accumulating) or harmful (faster clearing gives borrowers less time to add collateral).

Check the composition of any Curve or Uniswap pool that you provide liquidity. If a stablecoin has a disproportionately high share of the pool, the pool has been priced at moderate unanchor risk and your exposure to volatile losses is high.

Check whether the stablecoin issuer has released information about its reserve custodian. Circle discloses its banking relationships. Tether provides quarterly assurance reports but does not disclose specific custodians. The risk profile of unanchoring depends largely on the specific institution holding the reserves and its sensitivity to bank runs, regulatory action or operational failures.

Check your clearing thresholds. If you are borrowing stablecoin collateral, calculate how much the stablecoin price will need to fall for your position to be cleared. If your LTV is conservative, a 3% unanchoring lasting for 30 seconds may not trigger liquidation, but a 10% unanchoring will almost certainly trigger.

Check the redemption terms of stablecoins. Some stable coins can be redeemed 24/7. Others have processing windows, minimum redemption amounts or identity verification requirements, causing delays. These delays determine the speed at which the arbitrage resumes anchoring after an unanchoring event.

Directions to watch

Oracle infrastructure upgrades. Chainlink and other oracle providers are actively developing pull-based oracle models that allow protocols to request price updates on demand rather than waiting for push-based fixed schedule updates. These models will significantly reduce the oracle delay window during unanchoring.

Curve v2 is adopted with concentrated liquidity. A new AMM design that allows liquidity providers to concentrate capital within a specific price range may change the dynamics of unilateral liquidity during the unanchoring period, either reducing the slip point for large transactions or creating a cliff effect where liquidity disappears completely below a specific price.

Diversification of stablecoin reserves after relevant bills. Reserve requirements in the relevant bill may push issuers to move towards more diversified custody arrangements, reducing the concentration risk that could cause USDC to unanchor when SVB fails.

Cross-chain unanchored propagation. As stablecoins bridge across multiple chains, unanchoring on Ethereum can spread to networks such as Arbitrum, Optimism, and Base, with latency depending on the bridging final time and oracle configuration on each chain.

Real-time clearing dashboard. Tools such as DefiLlama's Clearing Tracker provide real-time visibility of mortgage positions that will be cleared at different price levels. Monitoring these dashboards during periods of stablecoin pressure can provide early warning of potential clearing cascades.

What is a stablecoin de-anchoring?

Deanchoring of stablecoins refers to the phenomenon that the market price of a stablecoin deviates from its target value (usually US$1). The unanchoring may be triggered by external events such as an imbalance in exchange liquidity, concerns about issuer reserves, or the failure of a bank that affects custody of reserve assets. Most unanchors are short-lived and resolved through arbitrage, but severe unanchors can last for hours or even days.

How long does typical unanchoring last?

Most stablecoin price deviations last from seconds to minutes and are resolved by automated arbitrage robots that buy discounted stablecoins and redeem or sell them at parity elsewhere. Severe unanchors caused by solvency issues (such as the USDC during the SVB collapse) can last for 48 hours or more because arbitrageurs are reluctant to buy until they confirm that the redemption mechanism is operating properly.

Will 30 seconds of unanchoring cause actual damage?

Yes. DeFi loan agreements use oracle pricing that is updated at regular intervals. When the oracle updates and reflects lower stablecoin prices, positions previously above the clearing threshold may suddenly become eligible for clearing. The liquidator confiscated the collateral at a discounted price, and the borrower lost part of his funds. This can happen within a oracle update cycle.

What role does the oracle play during unanchoring?

Oracle provides price data used by the DeFi protocol to evaluate collateral and determine eligibility for liquidation. The oracle price of most stablecoins is updated after a price change of more than 0.25% or after a fixed time interval. This creates a lag between the market price and the agreement view price, which may delay or accelerate liquidation depending on the timing.

Why does the Curve pool amplify and unanchor?

Curve's stable exchange design concentrates liquidity around the 1:1 price ratio, thereby minimizing slip points in normal trading. During the unanchored period, sellers sell unanchored assets into the pool, changing its composition. As the pool becomes increasingly one-sided, implied exchange rate nonlinearities deteriorate, making it increasingly costly for arbitrageurs to restore balance.

What is the difference between liquidity unanchoring and solvency unanchoring?

Liquidity unanchoring occurs when selling pressure temporarily exceeds the buying capacity of the secondary market, but the issuer's reserves are intact and the redemption mechanism is operating normally. Such unanchoring is usually resolved within minutes by arbitrage. Solvency unanchoring occurs when the issuer's reserves are insufficient to pay all redemptions at par, which may lead to continued price declines and potential permanent losses.

How can an arbitrage robot restore its anchoring?

Arbitrage robots monitor price differences between different locations. When stablecoins are traded under $1 on DEX and at parity on CEX, the robot buys on DEX and sells on CEX. If the redemption mechanism is normal, the robot can also buy discounted stablecoins and directly redeem them from the issuer for US$1. This buying pressure pushed DEX prices back to parity.

How can users protect themselves during unanchoring?

Users can reduce exposure by maintaining a conservative loan-to-value ratio when borrowing stablecoin collateral, decentralizing allocations across multiple stablecoin issuers, monitoring oracle update plans for the protocols used, and checking disclosures from reserve custodians holding stablecoin. Avoiding concentrating a single stablecoin in the liquidity pool can also reduce the risk of erratic losses in the event of unanchoring.