Stablecoin failures, algorithmic unwinds, reserve/collateral shenanigans (USD1, Tether), and market manipulation
Built 2026-06-12 from research/macro-stablecoin-failures-manipulation.json. On-chain claims verified this pass against Arkham Intelligence (via Forbes / DL News) and Etherscan; adjudicated cases from SEC/CFTC/DOJ/NYAG releases.
Grade discipline. Algorithmic unwinds are hard facts (on-chain, dated, billions wiped). Tether's reserve misrepresentation is adjudicated for the past (NYAG 2021, CFTC 2021); current full-backing rests on attestation, not audit. The USD1/WLFI collateral concentration is on-chain-documented fact (Arkham); characterizing it as engineered conflict is labeled interpretation. The Tether-pumps-Bitcoin thesis is a contested academic finding, not proven causation. Overlay edges excluded from the proofs.
1. Algorithmic stablecoins — a repeated, predictable failure mode
- TerraUSD (UST) / Luna — May 2022. UST held $1 only by mint/burn arbitrage against LUNA; Anchor's unsustainable ~20% yield concentrated reflexive demand. An outflow broke the peg, the arbitrage hyperinflated LUNA, and ~$40–60B evaporated in days. Triggered the 2022 contagion (Three Arrows, Celsius, Voyager). Do Kwon charged by SEC + DOJ, arrested in Montenegro; Terraform settled with the SEC for ~$4.5B (2024). Fact.
- Iron Finance (TITAN) — Jun 2021. Partial-algo; a bank-run dynamic took TITAN ~$60→~$0 in a day — "the first large-scale crypto bank run." Mark Cuban burned, then called for regulation. Fact.
- Basis Cash / Empty Set Dollar (2020–21). A wave of seigniorage-style algos (Basis Cash co-founded pseudonymously by Do Kwon) that all de-pegged — the failure mode was known before Terra, which was a scaling of it, not a surprise. Fact.
- USDD (Tron) — 2022–. Justin Sun's hybrid; lost peg in 2022 and traded below $1 for extended periods, propped by TRON DAO reserves. A live persistently-under-pegged "stablecoin." Fact.
- USDN/Waves, DEI, Fei, Beanstalk (2022). Further de-pegs and a flash-loan governance drain (Beanstalk). The category's failure rate is the point. Fact.
2. Reserve / collateral shenanigans
Tether (USDT):
- NYAG settlement (Feb 2021, $18.5M): Tether/Bitfinex misrepresented full backing; Bitfinex covered an ~$850M shortfall with Tether reserves. Adjudicated: not always fully cash-backed. Fact.
- CFTC fine (Oct 2021, $41M): USDT fully backed only part of the time over 2016–18. Fact.
- Reserve shift: commercial paper → US Treasuries; now a notable marginal T-bill buyer (the stablecoin-Treasury rail), but on attestation, not full audit. Fact (attestation) / weak (no audit).
World Liberty Financial / USD1 — on-chain-verified this pass:
- Structure & contract. WLFI is a Trump-family-interest DeFi venture; USD1 launched Mar 2025 on Ethereum + BNB Chain. Ethereum contract
0x8d0d000ee44948fc98c9b98a4fa4921476f08b0d(~44k holders, Jun 2026); BitGo Trust custody; backed (per issuer) by cash + short T-bills via government MMFs. A sitting president's family profiting from a dollar token while the administration sets stablecoin policy (GENIUS Act) is an unprecedented direct conflict. Fact. - MGX / Binance $2B settlement (on-chain). Abu Dhabi's MGX invested ~$2B into Binance settled in USD1 (announced at Token2049, May 1 2025, by WLFI co-founder Zach Witkoff — son of envoy Steve Witkoff) — the largest investment ever paid in stablecoins. The mint/transfer are on-chain; the deal placed a large share of USD1 reserves under Binance's custody and generated reserve interest income for WLFI. Fact (on-chain).
- The documented collateral/concentration anomaly (the documented point — verified). Per Arkham Intelligence (Forbes Feb-2026; DL News), Binance custodies ~87–89% of all USD1 in circulation — roughly $4.7B of a ~$5.4B supply. A single exchange holds nearly the entire float of the president's-family stablecoin, into which a $2B Gulf-sovereign flow was routed, with interest accruing to WLFI. The concentration is the on-chain fact; "engineered conflict" is the labeled interpretation. Fact (Arkham on-chain).
- Cantor distribution. USD1 expanded into the Cantor (Lutnick-linked) network — tying distribution to a Commerce-Secretary-linked firm. Fact (reported; on-chain-verifiable).
- CZ pardon proximity. Changpeng Zhao was pardoned (2025) while Binance custodied ~88% of USD1 and had taken the MGX settlement. The sequence is documented; causal corruption is alleged, not adjudicated — left contested. Fact (sequence) + labeled interpretation.
USDC / Circle (the bank-link): $3.3B of reserves at SVB; the bank's failure depegged USDC to ~$0.87 (Mar 2023) until the FDIC backstop. A "fully reserved" coin broke because the reserve bank broke — the link to spec-crypto-banking-debanking. Fact.
3. The manipulation layer
- Tether-prints-pump-Bitcoin. Griffin & Shams (J. Finance 2020) found USDT issuance timed with BTC support in 2017, consistent with single-actor manipulation — influential but contested (timing/correlation, not a verdict). Contested.
- Wash trading. Studies/enforcement find large fractions of reported crypto volume are wash trades — inflating apparent depth. Fact (pattern).
- Alameda/FTX. A dominant market-maker that used FTX customer funds (SBF convicted Nov 2023). The lesson: one actor as exchange + market-maker + token issuer is the manipulation-enabling configuration. Fact (convicted).
4. Custody-concentration monitor (added 2026-06-16, #65)
Generalizing the cases into a monitoring framework — across every documented failure the common variable is concentration, so tracking these six metrics makes the danger legible before the peg breaks. Current readings fact (on-chain/attestation, Jun-2026); thresholds a labeled model.
| Metric | Current reading | Danger threshold | What it flags |
|---|---|---|---|
| Float concentration (1 venue's % of supply) | USD1: Binance ~87–89% (~$4.7B/$5.4B) | >50% = single point of failure; >80% = the token is that venue's instrument | a venue freeze/failure breaks the peg or traps the float |
| Reserve-custodian concentration | USDC had $3.3B at SVB → depeg; USD1 via BitGo + gov MMFs | any single uninsured bank holding a material share | the SVB lesson — a "fully-reserved" coin breaks if the reserve bank breaks |
| Issuer concentration of the Treasury bid | Tether >$100B (~18th-largest US-debt holder) | one issuer > sovereign-scale share of the marginal bid | a single-issuer run becomes a Treasury-market event |
| Audit vs attestation | Tether: attestation, no full audit | no full audit on a sovereign-scale issuer | NYAG/CFTC precedent — "fully backed" was adjudicated false for the past |
| Redemption-access | par redemption ~ authorized participants only | narrow AP set + concentrated float | retail can't redeem at par in a run — defended by few, broken by many |
| Chain concentration | most supply on 1–2 chains | supermajority on one chain | a chain halt/exploit freezes that share |
Synthesis: USDC/SVB (custodian), Terra (single mechanism), USD1 (87–89% float at Binance), Tether (single-issuer bid + attestation-only) — concentration converts an idiosyncratic counterparty failure into a peg break. A custody-concentration monitor is the cheapest early-warning gauge. USD1 is the live red flag: a president's-family token whose float is ~88% at one pardoned-founder exchange, fed a $2B Gulf-sovereign flow, with reserve interest accruing to the issuer.
4b. Honest limits
Algorithmic unwinds and adjudicated cases (Terra/Do Kwon, Tether NYAG/CFTC, FTX/SBF) are hard facts. The USD1/WLFI collateral concentration is on-chain-documented (Arkham), not speculative — verified this pass. The corrupt-intent characterization (incl. the CZ-pardon proximity) is labeled interpretation, left contested rather than asserted. Current Tether full-backing rests on attestation, not audit. Do Kwon and Justin Sun are added to the Persons tab; SBF and the Gulf/Trump/Lutnick nodes already exist.
Sources for the on-chain verification: Arkham Intelligence via Forbes/crypto.news and DL News; contract on Etherscan; deal confirmation Cointelegraph / Fortune.
← Research index · structured data: macro-stablecoin-failures-manipulation.json · macro-stablecoin-failures-manipulation.md