Theta
Memecoin Margin & Short Platform on Robinhood Chain — Build Research
Research snapshot: 14 September 2026. Three parallel research tracks (chain infrastructure, protocol architecture, market + regulatory) were run and synthesized here. Items marked [unconfirmed] could not be corroborated from a primary source.
0. Verdict
Build an oracle-priced, vault-backed perpetuals engine with isolated per-token markets, not a borrow-and-sell margin system. This is the only design that can offer shorts on the memecoin long tail. Real "borrow the token and sell it" shorting (the tradfi analogue) is structurally unworkable for memecoins: nobody lends them, borrow rates spike to hundreds of percent, and liquidators cannot buy enough spot to cover a losing short. Every venue that lists memecoin leverage today (Hyperliquid, Drift, Aster, Gemini) uses synthetic perps at 3–5x, not borrowed tokens.
The hard part is not the margin math. It is the oracle and the size-vs-liquidity limits. Every major memecoin-leverage blowup in 2025 (JELLY, XPL, POPCAT on Hyperliquid) was a case where a position larger than the spot market could absorb got dumped onto the backstop vault. The one big 2026 oracle loss (Ostium, $23.75M) was a stolen signer key, not on-chain logic.
Robinhood Chain is ready for this. Mainnet since 1 July 2026, permissionless deployment, Uniswap v4 as the dominant AMM, Morpho and Chainlink live, USDG as the native stablecoin, and a memecoin launchpad scene (Pons, Long.xyz, Pools.trade) that already produces ~80% of the chain's DEX volume. Nobody on the chain offers memecoin shorts yet. Lighter and Arcus do perps, but on majors, commodities and stock tokens.
The regulatory shape is fixed: memecoins are "digital collectibles" (not securities), but leveraged retail trading on them is a CFTC futures matter in the US. Every Robinhood-touched leverage product excludes US persons. Plan for an offshore entity, a geo-blocked front-end, and permissionless contracts from day one.
1. Robinhood Chain: what you are building on
| Item | Fact | Source |
|---|---|---|
| Status | Mainnet live 1 Jul 2026; testnet since Feb 2026, still running | Robinhood newsroom |
| Chain IDs | Mainnet 4663, testnet 46630 (settles to Sepolia) | docs.robinhood.com/chain |
| Stack | Arbitrum Nitro (Orbit), ArbOS 61, settles to Ethereum L1 as a rollup with blob DA. Not AnyTrust. | Arbitrum DAO factsheet, L2BEAT |
| Blocks | ~100 ms, FCFS ordering, no priority auction | docs |
| Gas token | ETH. No chain token exists. | docs |
| Sequencer | Single, run by Robinhood. Has a transaction-filtering precompile. 13-min halt on 4 Sep 2026, cause undisclosed. | docs, CryptoSlate |
| Governance | 8-seat Security Council (Robinhood x2, BitGo, Chainlink, Fireblocks, Offchain Labs, Paxos, Talos). 7-day timelock on routine upgrades. | docs/governance |
| Deployment | Permissionless. Foundry/Hardhat, standard EVM. Stylus (Rust) confirmed on testnet, very likely on mainnet. | docs, Arbitrum Foundation |
| Stablecoin | USDG (Paxos) is native and dominant (~68% of stablecoin cap). USDe second. USDC is not natively issued by Circle; Across converts inbound USDC to USDG. | Circle chain list, Across blog |
| Oracles | Chainlink only in official docs: Data Feeds (crypto majors + every stock token), Data Streams, Sequencer Uptime Feed, CCIP. No Pyth/RedStone/Chronicle deployment found. No memecoin price feeds exist. | docs/oracles, Chainlink |
| DEXs | Uniswap v2/v3/v4 + UniswapX (~$200M TVL, >50% of all Uniswap v4 volume), Arcus (RFQ+AMM, RWA perps to 50x), Rialto, Pleiades, 1inch/0x. PancakeSwap/Balancer/Aerodrome reported [unconfirmed]. | Uniswap blog, DefiLlama |
| Lending | Morpho Blue (backbone of Robinhood Earn, ~$900M deposits, Steakhouse-curated). No Aave/Euler/Compound. | Morpho blog |
| Perps | Lighter (zk L3 on Robinhood Chain, in-app perps, stock tokens as margin, US-blocked), Arcus (RWA perps), LongX (3x NVDA wrapper). | The Block, L2BEAT |
| Memecoins | Native, launchpad-minted: Noxa (dead, ~60k tokens), Pons (dominant, graduates to Uniswap v4), Long.xyz (memecoins paired against stock tokens like NVDA), Pools.trade (Uniswap's own launchpad), CASHCAT ($230M peak, listed on Robinhood app and Hyperliquid at 3x). Memecoins = ~79% of DEX volume. | CoinDesk, CoinGecko, The Defiant |
| Volume | Record $1.6B/day DEX volume (1 Sep 2026). Lighter perps $354M/day. Caveat: 90-day gas subsidy ends ~29 Sep 2026. | Crypto Briefing, KuCoin |
| Infra | Alchemy/QuickNode RPC, Blockscout explorer, The Graph + Goldsky indexing, ERC-4337 v0.6/0.7/0.8 + EIP-7702, Alchemy/ZeroDev paymasters, Privy/Dynamic embedded wallets, Rabby/MetaMask/Trust Wallet support. | docs |
| Bridges | Canonical Arbitrum bridge (7-day exit), LayerZero/Stargate, CCIP, Across, Relay, LiFi. | docs/bridging |
| Grants | No dedicated program. $1M Robinhood sponsorship of Arbitrum Open House 2026 buildathons (Singapore starts 14 Sep 2026). | Arbitrum Foundation |
| ToS | Third-party dApps unrestricted; no derivatives/leverage prohibition; VPN evasion prohibited; Robinhood "does not control what third parties build." | docs/terms-of-service |
Implications - Collateral asset should be USDG, not USDC. - You must build your own memecoin oracle. Chainlink will not have feeds for Pons launchpad tokens. - Stock-paired memecoins (Long.xyz) price through a stock token, so the oracle has to compose memecoin/NVDA pool price × Chainlink NVDA feed. - Morpho Blue is already there if you later want physically-settled shorts on the top few tokens. - Sub-second blocks mean keeper latency matters less than on Arbitrum One, but a centralized sequencer means you need the Chainlink Sequencer Uptime Feed in your liquidation logic (grace period after downtime).
2. What "margin and short" means on-chain
Tradfi margin: the broker lends you cash (long) or borrows shares from another client and sells them for you (short). On-chain there are three ways to reproduce that:
| Model | What the user holds | Counterparty | Needs token lenders? | Examples | Fit for memecoins |
|---|---|---|---|---|---|
| A. Oracle-priced vault perps | Synthetic PnL claim | LP vault | No | GMX v2, Jupiter Perps, gTrade, Ostium | Best. Works for any token with a defensible price. Bounded by vault size. |
| B. Orderbook perps | Synthetic PnL claim | Other traders + backstop vault | No | Hyperliquid, dYdX v4, Lighter, Drift | Best UX for liquid names; needs matching engine and market makers. Thin books on long tail end up vault-backed anyway. |
| C. Real borrow-and-sell | Real debt in the token | Lenders of the token | Yes | Aave, Morpho, Euler, Kamino; Contango/Gearbox on top | Only viable for top ~10 memecoins with curated lending markets. Not viable for the long tail. |
| vAMM | Synthetic | Virtual curve | No | Perp Protocol v1/v2 | Dead. Being sunset. |
Why C fails for memecoins (this is what "just like stocks" would literally mean): 1. No supply: someone must lend the memecoin and eat the bad debt. 2. Borrow rates blow past 200% APR when shorts crowd in. 3. Positions and liquidation prices are public; communities coordinate squeezes. 4. Liquidators must buy spot to repay; if pool depth is $200k and the short is $2M, it becomes bad debt. 5. The debt token is the illiquid asset, so a TWAP pump moves the liability directly.
Recommended: Model A with isolated per-market exposure, following the GMX v2 contract topology. Optional Phase 3 hybrid: Morpho Blue isolated markets for physically-settled shorts on the top 3–5 tokens, and an internal orderbook for majors if volume justifies it.
3. What you need to build
3.1 Smart contracts (Solidity, EVM)
| Module | Responsibility | Reference design |
|---|---|---|
| Router / Registry | Single entry point; service discovery so modules can be repointed | Ostium IOstiumRegistry, GMX ExchangeRouter |
| DataStore + Vault | All state and all funds, separated from logic so logic can be upgraded | GMX v2 DataStore/Bank pattern |
| OrderHandler | Two-step flow: user submits order on-chain → keeper executes in a later tx bundling signed oracle prices. Prevents front-running of oracle updates. | GMX v2, gTrade, Ostium |
| PositionManager / Clearinghouse | Margin math, isolated margin per market, entry/exit, PnL settlement in USDG | Hyperliquid strictIsolated, Lighter AllocatedMargin |
| Oracle | Verify signed price reports; staleness, deviation, per-update move clamps, sequencer-uptime check | See 3.2 |
| FeeModule | Open/close fee, price-impact fee (imbalance-based, capped per market), funding (dominant side pays, clamped), borrowing fee rising with OI/maxOI | GMX v2 formulas, gTrade 1%-depth impact |
| RiskConfig | Per-market: max leverage, initial/maintenance margin, OI cap per side, per-account cap, impact caps, funding clamp. Timelocked changes. | HIP-3 deployer params |
| LiquidationModule | Keeper-triggered; partial liquidation with throttle; try SL/limit before full liquidation; liquidation fee | Hyperliquid 20%-chunk partials, Euler cool-off |
| Backstop + InsuranceFund + ADL | Waterfall: partial liq → backstop vault up to a per-market budget → insurance fund → auto-deleverage profitable positions ranked by uPnL × leverage → halt and settle at mark | Lighter LLP, HIP-3 haltTrading |
| LPVault (ERC-4626) | USDG vault as counterparty; reserve factor, pending-PnL caps on deposit/withdraw, optional isolated sub-vaults per market | GMX GM pools, Arqen isolated pools, Ostium buffer/OLP two-tier |
| Pause / Guardian | Emergency pause, per-market halt, timelock for anything touching collateral or oracle config | Lesson from Drift (Apr 2026, $270M, zero-timelock council migration) |
| Listing module | Permissionless or curated listing gated by measurable spot-depth thresholds | Section 4 |
Upgradeability: transparent/UUPS or beacon proxies + timelock + guardian. Keep the vault near-immutable and the risk config governed.
3.2 Oracle (the core engineering problem)
No external feed exists for launchpad memecoins. You will build a composite:
- On-chain TWAP/TWMP from Uniswap v4 pools on Robinhood Chain. Use time-weighted median rather than mean, ≥2 pools where they exist, and check full-range liquidity. Consider the Uniswap v4 truncated oracle hook (caps tick movement per block at ±9,116 ticks, so a manipulated price takes ~15 blocks to propagate).
- Own signed index from an off-chain aggregator (your own DEX pool observations + any CEX/Hyperliquid listing for larger names), quorum-signed by multiple keys held in HSMs. The Ostium lesson: one compromised signer + forwarder drained $23.75M in 5 minutes.
- For stock-paired memecoins: pool price in NVDA-token terms × Chainlink NVDA feed, honoring
oraclePaused()during corporate actions. - Guards on-chain: staleness bound, deviation halt vs reference, per-update move clamp (HIP-3 uses 1% per update and 10x start-of-day cap), bid/ask-aware pricing (Euler-style spread that tightens effective leverage), Euler-style cool-off period after a health check to kill same-block flip attacks.
- Sequencer uptime feed integration: grace period on liquidations after sequencer downtime.
Cost-of-attack rule (Chaos Labs / Euler): never let the max attainable profit from a market exceed the cost of moving the index. Uniswap v3 TWAP manipulation cost over a 30-minute window is astronomical for 1 block but only ~2x price for 72 consecutive blocks; on a single-sequencer chain with 100 ms blocks, assume an attacker can hold price for many blocks cheaply, so weight the median/clamps heavily.
3.3 Off-chain services
| Service | Notes |
|---|---|
| Price aggregator + signer | Polls pools, computes index, signs with quorum. Two independent operators minimum. |
| Order-execution keepers | Bundle signed prices and execute pending orders. Permissionless with fee, plus your own fleet. |
| Liquidation keepers | Enumerate positions on-chain (GMX exposes enumerable sets so keepers don't rely on indexer lag). Run ≥2 operators. |
| ADL / funding cranks | Accrue funding and borrowing; trigger ADL when backstop budget breached. |
| Indexer | Goldsky or The Graph (both support Robinhood Chain) for history, PnL, leaderboard. |
| WebSocket feed | Real-time prices, positions, funding to the frontend. |
| Risk monitoring | OI vs spot-depth dashboards, related-wallet clustering (POPCAT attack used 19 wallets), auto-halt when mark deviates from external index. |
| Frontend | Geo-blocked, ToS-gated, wallet connect (Rabby/MetaMask/Robinhood Wallet), ERC-4337 for gasless UX via Alchemy/ZeroDev paymaster. Long/short in one click. |
3.4 Security and operations
- ≥2 independent audits (budget $60–120k each round in 2026 pricing) + a contest (Sherlock/Cantina/Code4rena) + bug bounty before real TVL.
- Key management: HSM or MPC for oracle signers; no pre-signed governance txs; timelocks on collateral whitelisting.
- Chain-halt runbook (dYdX halted 8h in Oct 2025; Robinhood Chain halted 13 min in Sep 2026).
- Publish force-settlement rules in advance. Hyperliquid's discretionary $0.0095 JELLY settlement was its biggest reputational hit.
4. Listing and risk rules for memecoins
Industry practice, from published parameters:
| Parameter | Observed practice | Suggested default for a new venue |
|---|---|---|
| Max leverage, liquid memecoins (DOGE, PEPE, WIF class) | Hyperliquid 10x up to $20M notional, 5x above; Drift 5x; Aster 5x; Gemini EU 100x | 5x |
| Max leverage, fresh launchpad tokens | Hyperliquid 3x on TRUMP at launch, 3x on CASHCAT; Hyperps 3x | 2–3x |
| Maintenance margin | Hyperliquid: half of initial (5% at 10x) | Half of initial |
| OI cap per market | HL dynamic from liquidity/basis/leverage; HIP-3 mandatory notional caps; Arqen 10% of pool per token | ≤10–25% of 2%-depth spot liquidity, recomputed daily |
| Per-account cap | Not always published; JELLY passed HL's dynamic cap with $4M | Fraction of market OI cap |
| Price impact | GMX: imbalance^exp × factor, caps 75–1,000 bps for thin markets; gTrade: 1%-depth linear | GMX-style with 100–1,000 bps caps |
| Funding | GMX floor 15% / cap 300% APR; HIP-3 clamp default 0.0003 per interval | Clamped, dominant side pays |
| Borrowing fee | gTrade base × (OI/maxOI)^exp; GMX kink ~50% APR at 75% util |
Rises with OI/maxOI |
| Liquidation fee | GMX 0.20–0.45% of size (0.45% on new volatile listings); dYdX ≤1.5% | 0.3–0.5% |
| Margin mode | HIP-3 strictIsolated; Drift pre-launch forces isolation | Isolated only at launch |
| Collateral | USDC-only at GMX/Ostium/trade.xyz | USDG only at launch |
| Listing threshold | Kalshi/CFTC: "deep, active, continuous spot market" | Minimum pool age, minimum full-range liquidity, cost-to-move-index ≥ N × OI cap, no single holder >X% |
| Delisting | HIP-3 haltTrading settles at mark |
Pre-published trigger + settlement methodology |
2025–26 incidents to design against
| Incident | What happened | Design fix |
|---|---|---|
| JELLY, Hyperliquid, Mar 2025 | $4M short self-liquidated into the backstop vault while attacker pumped spot; HLP −$13.5M unrealized; validators force-settled | Backstop budget per market; related-account detection; pre-published settlement rules |
| XPL, Hyperliquid, Aug 2025 | Whales bought entire ask side, +200% in 2 min, $130M liquidated | Self-referential oracles are squeezable; per-update clamps + OI vs depth |
| POPCAT, Hyperliquid, Nov 2025 | 19 wallets, $26M of 5x longs, spoofed $20M wall pulled; $4.5–4.9M bad debt | Wallet clustering; per-account caps; mark smoothing |
| Ostium, Jul 2026 | Stolen oracle signer + forwarder key; $23.75M drained in 5 min | Quorum signing, HSM keys, on-chain deviation bounds vs independent feed |
| Drift, Apr 2026 | DPRK social engineering; zero-timelock council migration; fake collateral whitelisted; $270M+ | Timelocks on collateral and admin; no durable-nonce pre-signed txs |
| dYdX, Oct 2025 | Liquidation accounting bug halted chain 8h | Fuzz/invariant tests on liquidation math |
| 10 Oct 2025 cascade | $19B liquidations, cross-margin propagated losses; memecoin cap −40% by year-end | Isolated margin; circuit breakers |
5. Regulatory shape and entity structure
Asset classification - SEC staff (27 Feb 2025) and the SEC/CFTC joint interpretation (17 Mar 2026): memecoins are "digital collectibles," not securities. DOGE and SHIB are named digital commodities. - Derivatives and leverage on them fall under the CFTC regardless.
US leverage rule - CEA §2(c)(2)(D): retail leveraged/margined commodity trades are futures unless "actual delivery" within 28 days. The CFTC withdrew its 2020 actual-delivery guidance in Dec 2025 and has not replaced it. - Uniswap Labs was fined $175k (Sep 2024) for running a front-end to leveraged tokens. The theory targets interface operators, not contracts. - Perps are coming onshore only via registered DCMs (Coinbase, Bitnomial, Kalshi BTCPERP approved May 2026, limited to assets with "deep, active, continuous" spot markets). Nothing permits a DeFi perp protocol to serve US retail. CFTC Chair Selig has promised guidance on when software providers must register; not issued as of 14 Sep 2026. - CLARITY Act: Senate cloture vote 15 Sep 2026; ~25% odds of enactment in 2026; does not address perps or memecoins.
Developer liability - DOJ Blanche memo (Apr 2025): no prosecution of developers for users' conduct. But Samourai founders were sentenced (Nov 2025) and Roman Storm's §1960 conviction stands with a retrial set for Apr 2027. Non-custodial design matters; BRCA (S. 3611) would formalize the "control" safe harbor but is still at introduction.
Robinhood-specific - Chain ToS permits third-party DeFi; no leverage prohibition. But Robinhood excludes US users from every leverage product it touches, runs the sole sequencer with a filtering precompile, and is a US broker-dealer/FCM. Whether it would filter a US-facing leverage dApp is unknown; no policy published. - EU: Robinhood's own perps (Bitstamp, MiFID II) compete with any EU-facing on-chain product; an EU-facing front-end likely needs licensing or a non-EU structure.
Practical structure used by peers - Contracts: permissionless, immutable-leaning, non-custodial. - Interface: separate offshore entity (Hyperliquid Corp., dYdX Operations Services Ltd. pattern), IP geo-block for US/UK/CA/CH/UAE/SG, ToS with VPN ban, close-only mode for flagged wallets. - Entity: Cayman foundation + BVI token issuer + Panama/BVI front-end company is the common stack. - Optional: zk-KYC (ZKPassport, Privado ID) for jurisdiction proofs without identity; ECP-only tier for US institutions.
6. Landscape and the gap
| Product | Memecoin shorts? | Where | Leverage | Gap it leaves |
|---|---|---|---|---|
| Hyperliquid core | Top ~30 memecoins only | Own L1 | 3–10x | Nothing below ~$100M cap; US blocked |
| Hyperliquid HIP-3 | Any deployer can list; none target memecoins | Own L1 | 1–50x, deployer sets | 500k HYPE stake (~$20M+) to deploy |
| Drift | ~30 memecoins | Solana | 5x | Post-$270M exploit; Solana only |
| Jupiter Perps | None | Solana | — | — |
| Lighter (in Robinhood Wallet) | Not confirmed | Robinhood Chain L3 | 10x | Majors/commodities/stocks; US blocked |
| Arcus | None | Robinhood Chain | 50x RWA | RWA focus |
| Aster / Bybit | PONS, CASHCAT class only | Own L1 / CEX | 5–20x | Off-chain |
| dumpy.fun / IMF | Borrow-and-sell, handful of tokens | Solana / Ethereum | ~1x | Tiny, dormant [unconfirmed] |
| Polymarket | Binary markets on memecoins | Polygon | — | No continuous exposure |
| Axiom / BullX / GMGN | Route to Hyperliquid perps | Terminals | up to 50x | Distribution only |
The gap: no venue offers shorts on Robinhood Chain-native launchpad tokens (Pons, Long.xyz, Pools.trade graduates), and none anywhere serves the long tail below Hyperliquid's listing bar. Memecoin/stock-token pairs are unique to Robinhood Chain and unpriceable by anyone else's oracle.
Demand signals: TRUMP launch drove Hyperliquid to $21–22B/day; FARTCOIN OI hit $1B; CASHCAT did $98M/day spot on Robinhood Chain and got a 3x perp on Hyperliquid within weeks. Counter-signal: demand is highly cyclical, and on 14 Sep 2026 DOGE was the only memecoin in Hyperliquid's top pairs.
7. Open questions to settle before building
- Oracle proof-of-concept first. Can you produce a manipulation-resistant index for a Pons-graduated token from Uniswap v4 pools alone? Measure cost-to-move for the top 20 Robinhood Chain memecoins.
- Who seeds the vault? LP vault needs $1–5M USDG to make 3x shorts meaningful. Is it protocol-owned, community, or one backer?
- Listing policy: curated (you pick 10–20 tokens) vs permissionless with thresholds (Arqen model). Curated is safer for launch.
- Robinhood's tolerance: get a read from the Robinhood Chain team (Arbitrum Open House Singapore starts 14 Sep 2026) on whether a leverage dApp that geo-blocks the US would be filtered.
- Entity and jurisdiction before any front-end goes live. Which countries do you block?
- Stylus or Solidity? Stylus (Rust) could run heavier oracle math cheaply, but tooling/audit cost is higher. Solidity is the safe default.
- Fork or write? GMX v2 synthetics is open-source and the closest template; Ostium contracts are public too. Forking cuts months but inherits their assumptions (Chainlink pricing).
8. Suggested phasing (not started)
| Phase | Scope | Exit criterion |
|---|---|---|
| 0. Oracle research | Index for top 20 Robinhood Chain memecoins; manipulation cost model; stock-paired composition | Cost-to-move ≥ 10× intended OI cap for each listed token |
| 1. Testnet perps | GMX-style vault perps on chain 46630; 5 curated tokens; USDG collateral; 3x; isolated; keepers | Invariant tests pass; simulated JELLY/POPCAT attacks fail |
| 2. Audited mainnet, curated | 2 audits + contest; $1–5M vault; 10–20 tokens; geo-blocked front-end; offshore entity | 30 days without bad debt |
| 3. Expand | Permissionless listing with thresholds; Morpho Blue physical shorts for top tokens; optional internal book for majors; multi-asset collateral with haircuts | Volume justifies MM program |
9. Key sources
Robinhood Chain: docs.robinhood.com/chain (about, connecting, oracles, bridging, governance, terms), forum.arbitrum.foundation factsheet, l2beat.com/layer2s/projects/lighter-robinhood, blog.uniswap.org/robinhood-chain-is-live, morpho.org/blog/robinhood-chooses-morpho, circle.com/multi-chain-usdc, coingecko.com/learn/robinhood-chain-built-for-rwa-loved-for-memes, coindesk.com (Noxa, AMC feud), thedefiant.io (Lighter collateral, Pons), theblock.co (mainnet, Arcus), chainstack.com/robinhood-chain-growth.
Architecture: docs.gmx.io, github.com/gmx-io/gmx-synthetics, hyperliquid.gitbook.io (HIP-3, liquidations, margin tiers, hyperps, price indices), docs.lighter.xyz, dydx.xyz/blog (v4 architecture, Oct 2025 incident), docs.morpho.org, github.com/euler-xyz/euler-vault-kit whitepaper, docs.gains.trade, docs.ostium.com, chaoslabs.xyz Uniswap v3 TWAP risk, github.com/euler-xyz/uni-v3-twap-manipulation, blog.uniswap.org/uniswap-v4-truncated-oracle-hook, halborn.com (JELLY, Ostium), arkm.com JELLY research, sherlock.xyz audit pricing.
Market & regulatory: sec.gov staff statement on meme coins (Feb 2025), SEC/CFTC joint interpretation (Mar 2026, via Ropes & Gray / Baker Botts), cftc.gov press releases 8961-24 (Uniswap), 9152-25, dechert.com CFTC perps (Jun 2026), coindesk.com/policy (Selig, Kalshi BTCPERP, Hyperliquid onshoring), justice.gov Samourai sentencing, defieducationfund.org Storm update and BRCA, help.dydx.trade geo-restrictions, datawallet.com Hyperliquid restrictions, soliduslabs.com pump.fun report, coingecko.com/learn/october-10-crypto-crash, elliptic.co Drift exploit, gemini.com memecoin perps, coinbase.com perps blog, bitstamp.net perps launch, robinhood.com/us/en/support/articles/robinhood-wallet-perpetual-futures.