The dollar peg is a mechanism, not a promise
Stablecoins compress several risks into one simple-looking number. A price of $1 may be supported by cash and Treasury bills, overcollateralised crypto, an issuer’s redemption window, market makers, an algorithm, or a mixture of those mechanisms. On a calm day, the differences are easy to ignore. During a run, they determine who can redeem, what can be sold, how fast liquidity disappears and whether the system can survive.
This article separates three ideas that are often blurred together: the exchange price, the redemption value offered by an issuer or protocol, and the actual value of the assets behind the token. Those numbers can diverge. A temporary exchange discount may close when redemptions work. A collateral shortfall may persist. An algorithmic design can enter a reflexive loop in which every attempted redemption weakens the asset meant to absorb it.
The charts below use historical one-hour candles from liquid spot pairs. They document what traders could actually see on the selected venue, not a reconstructed illustration. A low on one pair is not necessarily the lowest price on every exchange, and a market price is not the same thing as the issuer’s direct redemption price. That distinction is central to the investigation.
Four coins that shared a price target—but not the same risk
The label “stablecoin” says almost nothing about the machinery underneath. The first question is not whether a token normally trades at $1. It is what must remain true for the token to return to $1 under pressure.
| Asset | Peg design | Stress trigger | Captured market low | Outcome |
|---|---|---|---|---|
| UST | Algorithmic mint-and-burn with LUNA | Confidence loss, Anchor withdrawals and reflexive LUNA dilution | $0.225 on the captured UST/USDT window | Collapsed; the original peg never recovered |
| USDT | Issuer reserves plus direct redemption for eligible customers | Broad crypto panic and exchange demand for immediate liquidity | $0.909 on the captured USDT/BUSD proxy wick | Market price recovered while redemptions continued |
| USDC | Reserve-backed token with banking and Treasury exposure | Disclosure of $3.3B held at failed Silicon Valley Bank | $0.882 on the captured USDC/USDT window | Recovered after authorities protected bank depositors |
| DAI | Overcollateralised on-chain debt with material stablecoin collateral | USDC depeg transmitted through collateral and the Peg Stability Module | Below target across major venues | Recovered with USDC; design was later rebalanced |
The peg that printed its own escape route
TerraUSD did not promise a conventional reserve account containing one dollar for every UST. Its stability mechanism linked UST to LUNA: in theory, arbitrageurs could exchange one dollar of UST for one dollar of LUNA and restore the peg. That mechanism depended on LUNA retaining enough market value and on arbitrage capacity remaining usable when redemptions accelerated. Anchor Protocol added another dependency by attracting a very large pool of UST with a yield near 20%, making confidence in the peg, the yield and the Terra ecosystem mutually reinforcing.
When withdrawals and selling pressure pushed UST below $1, redemption created more LUNA. The additional supply weakened LUNA; a weaker LUNA required still more units to absorb the same UST; falling confidence produced more exits. This was not a normal reserve-backed redemption queue. It was a reflexive balance-sheet expansion during a run. Market depth vanished faster than the mechanism could restore trust, and the arbitrage that was supposed to close the gap began widening the system’s vulnerability.
Our UST/USDT capture starts on 7 May 2022 near parity and records an hourly low of $0.225 before the pair stopped producing candles in the selected window. The final economic damage went further: the SEC later said UST and related tokens fell close to zero and roughly $40 billion in market value was wiped out. In 2024 a US jury found Terraform Labs and Do Kwon liable for fraud, followed by a settlement exceeding $4.5 billion. The legal finding matters because the collapse was not merely described after the fact as a failed experiment; statements about stability were tested in court.
The central lesson is structural. A mechanism that works by issuing more of a volatile companion token is strongest when confidence is high and weakest precisely when mass redemptions begin. A stablecoin cannot be evaluated only by how efficiently it holds $1 in normal conditions. The relevant test is what the balance sheet becomes after ten, thirty or fifty percent of holders ask to leave at once.
UST/USDT · 1H
7–13 May 2022 · UTC
Drag, zoom and inspect the real hourly candles
A frightening market price did not equal an issuer default
As Terra collapsed, traders rushed between stablecoins, exchanges and bank-like off-ramps. USDT briefly traded below its target on several venues. Our chart uses the inverse of BUSDUSDT, a transparent proxy for the value of USDT in BUSD on the selected exchange, and captures a thin hourly wick near $0.909. That print is real, but it should not be misread as proof that every USDT traded at ninety-one cents or that all redemptions occurred at that value.
An exchange price reflects the urgency of buyers and sellers on that venue. Direct redemption is a separate route available under the issuer’s rules, minimums and compliance checks. If professional arbitrageurs can buy discounted tokens and redeem them near par, the gap creates an incentive to restore the price. That bridge can fail if reserves are impaired, banking access closes, redemptions are suspended or traders no longer trust the issuer. In May 2022, the market discount reversed rather than becoming a terminal collapse.
The episode therefore illustrates both resilience and opacity risk. A token may recover because the redemption mechanism functions, yet outsiders still depend on reserve reports, asset quality, banking access and operational capacity. “It came back” is evidence about that event, not a permanent guarantee. The useful question is whether the issuer can meet unusually large redemptions without selling illiquid assets at a loss or relying on a single banking channel.
USDT/BUSD proxy · 1H
11–14 May 2022 · UTC
Drag, zoom and inspect the real hourly candles
When a banking crisis entered a 24/7 market
USDC’s March 2023 depeg did not begin with a broken smart contract. It began when Silicon Valley Bank failed and Circle disclosed that $3.3 billion of USDC reserve cash was held there. Crypto markets trade through weekends, while traditional bank settlement and public information move on a different clock. Holders suddenly had to price a reserve asset whose immediate recoverability was uncertain, without waiting for banks to reopen.
The USDC/USDT chart captures the run in real time. Trading in the selected pair begins during the stress window and reaches an hourly low of $0.882. The discount then narrows sharply after US authorities announced that depositors at Silicon Valley Bank would be protected. The underlying token design had not changed during those hours; the market’s estimate of the reserve shortfall and the timing of access to cash had changed.
USDC recovered, but the event exposed a boundary that on-chain dashboards cannot remove. A reserve-backed token may settle on public blockchains while depending on commercial banks, custodians, Treasury markets and legal claims. Real-time token transfers do not make reserve assets real-time. Concentration at one bank, the treatment of uninsured deposits and the issuer’s ability to fund redemptions over a weekend can matter as much as smart-contract security.
This is why a reserve attestation must be read as a map, not a badge. Investors should look for the mix of cash and short-term securities, named custodians, maturity, concentration, publication lag and the legal entity that owes redemption. A one-to-one headline is only the start of due diligence.
USDC/USDT · 1H
11–14 March 2023 · UTC
Drag, zoom and inspect the real hourly candles
Decentralised branding, concentrated collateral
DAI was built around overcollateralised on-chain borrowing, but by 2023 a material part of its stability architecture depended directly or indirectly on USDC. The Peg Stability Module allowed swaps between DAI and approved stablecoins at tightly controlled prices. That made DAI more efficient in calm markets. It also meant that uncertainty about USDC could be transmitted into DAI even though the two tokens had different issuers and governance systems.
During the SVB weekend, DAI fell below $1 alongside USDC. The event demonstrated that decentralisation is not a binary property. A protocol may be governed on-chain and hold transparent collateral while still concentrating economic exposure in one fiat-backed token, one custodian class or one regulatory jurisdiction. Transparency allowed observers to see the link; it did not eliminate it.
DAI recovered as USDC recovered. Governance later reduced some of the concentration and changed the system’s balance-sheet strategy. The broader lesson applies to every collateralised stablecoin: list the assets that support it, then list what supports those assets. A basket of tokens can still represent one hidden risk if several components depend on the same bank, issuer or market maker.
Inside the first 24 hours of a depeg
A depeg usually begins before the chart visibly breaks. The trigger may be a bank failure, a large withdrawal, a hacked bridge or a disclosure about reserves, but the vulnerability was already embedded in the system. Concentrated deposits, subsidised yield, thin weekend liquidity, collateral made of related tokens or a narrow group of market makers can remain invisible while confidence is high. The first forensic task is therefore to separate trigger from cause. Silicon Valley Bank triggered the USDC shock; the deeper exposure was the concentration of reserve cash and the mismatch between a 24/7 token market and banking hours. Selling pressure triggered the UST break; the deeper problem was a reflexive redemption design attached to a volatile asset and an ecosystem dominated by yield-seeking deposits.
The second stage occurs in order books. Stablecoin markets do not have one global price. Every pair has its own inventory, market makers and withdrawal routes. If sellers urgently want USDT while buyers wait, USDC/USDT can fall even when the eventual dollar recovery is likely. If a venue suspends deposits or withdrawals, arbitrageurs cannot transport tokens to the market where the discount exists. A dramatic wick can therefore measure a temporary bottleneck rather than a final recovery value. This is why screenshots without the pair, venue, time zone and candle interval are poor evidence. Our charts label each of those elements and preserve the full observed window around the stress event.
The third stage tests the redemption bridge. Professional traders normally buy the cheaper token, redeem or swap it near $1 and keep the spread. That trade pulls the market back toward parity, but only if the bridge is large, fast and credible. Minimum redemption sizes, compliance checks, banking cut-off times, network congestion and uncertainty about reserve assets can reduce its capacity. A retail holder may be unable to use the direct route even when an institution can. The public market can therefore trade below $1 while the issuer continues honouring eligible redemptions at par. Conversely, an advertised redemption mechanism offers little protection if the asset paid out is itself collapsing, as happened when UST redemptions expanded LUNA supply.
The fourth stage spreads losses through collateral and leverage. A stablecoin used as collateral in lending protocols can trigger liquidations when its oracle price falls. Liquidity pools rebalance toward the weaker asset, leaving providers with more of the coin everyone wants to sell. Bridges may create additional versions whose redemption depends on another custodian or chain. Derivatives positions can be liquidated even if the peg later returns. DAI’s reaction to USDC illustrated this network effect: the problem did not have to originate inside DAI for DAI holders to feel it. Mapping only the token issuer misses the protocols, pools and credit positions built on top of it.
The fifth stage is a race between information and cash. Issuers publish statements, exchanges change limits, governance communities vote and authorities clarify the treatment of bank deposits. Markets price each update before accountants can produce a final answer. Good communication cannot repair missing reserves, but slow or ambiguous communication can deepen a temporary discount. The relevant evidence is operational: are redemptions processing, are reserve assets accessible, are transfers working, are counterparties identified, and can the system publish numbers that reconcile? A social-media promise without transaction capacity is not a stabilisation mechanism.
Finally, a recovered peg should not erase the incident. Analysts should record the lowest executable prices, duration below key thresholds, venue differences, redemption volume, changes in reserve composition and governance responses. Recovery demonstrates that a mechanism worked once under a specific shock. It does not prove that the next shock will arrive through the same channel. The most valuable output of a depeg is therefore not a prediction about which token fails next, but a dependency map showing what must keep working for each token to remain redeemable.
What the three real charts prove—and what they cannot prove
A peg can fail in different ways
UST shows a terminal reflexive collapse; USDT shows a temporary venue discount; USDC shows a reserve uncertainty shock. The same-looking move below $1 can represent very different balance sheets.
Liquidity is local
The lowest price depends on the venue, pair and hour. A dramatic wick may reflect thin order-book liquidity rather than the value available to every holder.
Recovery time is evidence
A fast return toward par suggests arbitrage and redemption were functioning, but it does not prove future solvency or eliminate concentration risk.
Contagion follows dependencies
DAI transmitted the USDC shock because collateral links connected them. Stablecoins should be mapped as a network, not reviewed one by one.
Eight checks before holding a stablecoin
- Identify the legal issuer or protocol and the exact redemption obligation.
- Separate the exchange price from the direct redemption price and eligibility rules.
- Read the reserve composition: cash, Treasury bills, loans, tokens and other assets.
- Check custodian and bank concentration instead of accepting a one-to-one headline.
- Look for liabilities, encumbrances and related-party exposure—not reserves alone.
- Map secondary dependencies: other stablecoins, bridges, market makers and protocols.
- Review what happens during weekends, banking holidays and a sudden redemption surge.
- Diversify operational cash and avoid treating any stablecoin as insured bank money.
The stablecoin test begins when everyone wants the same exit
UST, USDT, USDC and DAI did not experience the same event. One entered a death spiral and never recovered. One absorbed a severe exchange discount while direct redemption remained the key anchor. One imported a bank failure into a weekend crypto market. One inherited another stablecoin’s problem through collateral. Grouping them under a single “depeg” label hides the mechanism that determined the outcome.
The disciplined approach is to examine redemption, collateral, liquidity and legal claims before chasing yield. A token holding $1 today tells you that supply and demand are balanced today. It does not tell you what will happen when the market demands billions of dollars back at once.
Questions investors ask after a depeg
Does a price below $1 mean the stablecoin is insolvent?
Not automatically. It can reflect temporary exchange liquidity, restricted arbitrage or genuine reserve impairment. The redemption mechanism and collateral determine which explanation is plausible.
Is an algorithmic stablecoin always unsafe?
Designs differ, but a token backed mainly by endogenous assets can become reflexive under stress. The failure mode must be tested against mass redemptions, not average days.
Are reserve attestations the same as audits?
No. An attestation usually addresses specific information at a particular date. Scope, liabilities, asset ownership and the reporting entity still matter.
Can diversification across stablecoins remove the risk?
It can reduce single-issuer exposure only if the coins do not share the same banks, collateral, bridges or market makers. Hidden correlation can remain.
Risk note
Stablecoins can lose their target value, become illiquid or face redemption restrictions. This investigation is educational and does not recommend buying, selling or holding any token.
Sources and methodology
Market charts use unmodified hourly OHLC candles. Legal and structural statements are anchored to a small set of primary or institutional sources.
Binance Spot · USTUSDT · BUSDUSDT (1/x) · USDCUSDT · 1H · UTC