Derive (ex-Lyra) vs Deribit: pricing, settlement, collateral and the V3 migration
Data as of 26–27.09.2026 (exchange API snapshots 26.09.2026 ≈ 17:30 UTC) unless another date is given next to a figure.
Verdict. Derive prices options almost exactly like Deribit: settlement on a 30-minute average, and near-the-money marks on the two venues differ by 0.4–2.6 vol points. But Derive is the weaker counterparty and is about to migrate to a new version. Its yield-bearing tokens add nothing on top of Deribit: Deribit itself pays 3.40% on USDC, 4.75% on USDe and 2.25% on stETH held as collateral, while Derive pays 2.05% on USDC; DRV staking yields ≈ 1.5% a year, paid in DRV.
Derive's only extra value is what Deribit doesn't list: HYPE to 25.06.2027, ZEC to 26.03.2027, XAUT gold, ADA, LIT, PUMP, VVV, CC. But those series currently have no resting orders at all — the only way in is RFQ — and the price is a counterparty whose insurance fund equals 5.8% of borrowed funds, and which V3 will slice into four pieces (alts get ≈ 10%).
What it is
An on-chain exchange for European options, perps and spot on its own chain, Derive Chain (OP Stack, data on Celestia, operated by Conduit). Company: Lyra Technologies Corp, Panama; token DRV. On 13–14.09.2026 the team proposed V3 — a move to a ZK exchange settling on Ethereum L1 and shutting Derive Chain down; no migration date has been announced.
Sources: L2BEAT, Derive page, 27.09.2026 · forums.derive.xyz "DIP: Launch Derive V3" (14.09.2026) · docs.derive.xyz (the docs already describe V3, Sepolia testnet).
Findings
1. Yield-bearing collateral on Derive gives no edge over Deribit — Deribit pays more on the same collateral
| Collateral | Derive yield | Derive margin haircut (maintenance / initial) | Deribit yield on balance |
|---|---|---|---|
| USDC (idle cash) | 2.05% (lending pool) | 100% | 3.40% |
| USDe / sUSDe | 4.74% (accrues in sUSDe price) | sUSDe 80% / 70% | USDe 4.75% |
| wstETH / stETH | 2.2% (Lido, in price) | 80% / 72% | stETH 2.25% |
| weETH | 2.35% (in price) | 70% / 55% | — (not accepted) |
| DRV | staking ≈ 1.5% in DRV | 15% / 10% | — |
Collateral on Derive stays the holder's property inside the account, and a yield-bearing token keeps accruing. But for an options buyer that isn't extra money: premium is paid in dollars. Hold collateral in sUSDe or wstETH and borrow USDC against it, and the loan costs 3.92% APR — more than wstETH yields (2.2%) and only 0.8pp below sUSDe, while adding Ethena risk on top of venue risk. On Deribit the same USDe and stETH simply earn yield on balance.
DRV staking (stDRV). 1 DRV → 1 stDRV, non-transferable. Rewards: 100,000 stDRV per week shared across all stakers (since April 2026, down from 250,000), which the team itself put at ≈ 1.5% APR (was ≈ 3.8%). Unstaking takes 7 days, or is instant with a 20% haircut; 35% of protocol revenue goes to DRV buybacks. Fee discounts start at 200,000 stDRV (≈ $84k at $0.42 per DRV): option taker fee 0.015% instead of 0.03% — a $4 saving per $26,900 notional. For a buy-and-hold-to-expiry options buyer the discount is pocket change, while the DRV risk (−80% or worse is routine for exchange tokens) is full. Rewards paid in the same token are endogenous yield, not income from outside.
Sources: Derive API public/get_all_currencies and public/get_interest_rate_history, 26.09.2026 (USDC: supply 2.05%, borrow 3.92%) · Deribit API public/get_currencies, 26.09.2026 (apr field: USDC 3.40 · USDE 4.75 · STETH 2.25 · BUIDL 3.39) · DefiLlama yields, 26.09.2026 · help.derive.xyz "DRV" (23.04.2026) · forums.derive.xyz "DIP: Update DRV Buybacks, Staking Emissions and Unstake Period" (01.04.2026) · docs.derive.xyz "Institutional Trading Rewards Program".
2. Derive's option prices come from one vendor's surface, not its own book
Deribit marks options off its own vol surface built from its own order book, and settles on a 30-minute average of its index (07:30–08:00 UTC). Derive takes everything from a single source, Block Scholes: spot, forward, rate and the vol curve (a five-parameter SVI curve). The mark is the Black price off that curve, discounted at the risk-free rate. Expiry also uses a 30-minute spot average, at 08:00 UTC; over the final 30 minutes the mark glides from the forward to the already-printed average ("delta decay").
What it means for a long option: on standard margin a long option carries no margin requirement at all — an account holding only long options and no borrowing has nothing to liquidate. The mark only affects how much value the venue "sees" (relevant for borrowing against options, on portfolio margin only). The settlement price, though, is real money: it is set by Block Scholes signers, the chain needs both of two keys, and there is no independent check in the contract.
Mark comparison. Across a sample of December (25.12.2026) calls on BTC, ETH, SOL and HYPE, the two venues' marks agree within 0.4–2.6 vol points; the dollar differences mostly come from snapshot timing and different indices. The real difference is the book, not the mark: on a near-the-money ETH call the spread on Deribit is a few dollars, on Derive more than $500; on SOL and HYPE Derive has no orders at all. The way in is RFQ to market makers — at a price you don't know in advance.
Sources: docs.derive.xyz "Oracles", "Settlement", "Standard Margin", "Common Parameters" (V3, 27.09.2026) · Derive API public/get_ticker and Deribit public/ticker, 26.09.2026 ≈ 17:30 UTC.
3. Settlement: Derive is always USDC; Deribit settles in coin or via a future
On Derive an in-the-money option at expiry is "printed" as USDC in the account, with no settlement fee. On Deribit BTC and ETH options are inverse — paid in BTC or ETH, plus a delivery fee; linear USDC options (SOL, XRP, HYPE and others) have settled in two steps since April 2026 — first into a future at the strike, then into USDC. For a long option the only difference is the currency the payout arrives in. But on Derive a paid-out dollar is a claim on a USDC pool two-thirds of which is lent out — which is why it makes sense to withdraw the payout straight away.
Sources: docs.derive.xyz "Settlement", "Trading Fees" · Deribit API public/get_instrument (settlement_currency: ETH / BTC / USDC; fee 0.03%) · Deribit Support "Linear USDC Options".
4. V3 changes who covers losses — and for alts the fund gets ten times smaller
What V3 promises. Every state transition is verified by a ZK proof on Ethereum; data on Celestia; an "escape hatch" — the operator must process a withdrawal within two weeks or anyone can replace it. That is stronger than today's setup, but the proposal names no V3 audits.
The migration. 14 days' notice → bridges frozen 4h before → trading halted 5 min before → snapshot → new chain. Options on BTC, ETH, HYPE, SOL, ADA, XRP, ZEC, CC, LINK, DOGE, BNB, VVV, LIT, PUMP, XAUT carry over; anything not on the list is closed at mark in USDC. Some withdrawals after the move take hours to days, some another week via native bridges.
The insurance fund gets split. Of ≈ $3.84m, ≈ $0.30m is taken out and the rest allocated:
| Risk universe | Share | Amount |
|---|---|---|
| Prime (BTC, ETH) | 65% | ≈ $2.30m |
| HYPE | 20% | ≈ $0.71m |
| Alt | 10% | ≈ $0.35m |
| RWA (gold) | 5% | ≈ $0.18m |
A loss in one universe doesn't touch the others — but another universe's fund won't help either.
How losses beyond the fund are shared. V2: a temporary USDC withdrawal fee for everyone. V3: a one-off haircut on every solvent account in the universe, pro rata to its value; the whole universe is frozen while it runs.
Sources: forums.derive.xyz "DIP: Launch Derive V3" (14.09.2026) · docs.derive.xyz "Liquidations", "Risk Universes" · TradingView / KuCoin News, 14.09.2026 · on-chain SecurityModule $3.84m, 26.09.2026.
5. The rest of Derive's product set — and how much it weighs
- Options: European calls and puts in USDC. BTC and ETH to 24.09.2027 (same as Deribit); HYPE to 25.06.2027; ZEC to 26.03.2027; SOL, XRP, XAUT, ADA, LIT, PUMP, VVV, CC to 25.12.2026. Max tenor 400 days.
- Perps on ≈ 30 assets (BNB, DOGE, SUI, TAO, AVAX among them), zero-fee spot, USDC borrowing against collateral (3.92%); in V3, borrowing of coins themselves too (spot shorts without a perp).
- Vaults: basis trades, covered calls, principal-protected spreads. ≈ $1.74m in total: LBTC basis $0.84m (share price 1.122 since inception), weETH basis $0.73m (1.095), the rest under $75k each; three of ten are below 1.0. The API gives no inception dates, so annualized yields can't be established.
- RFQ and block trades: on multi-leg trades the second leg is fee-free; a straddle or vertical pays one fee.
- Fees: options taker $0.5 + 0.03% of notional (capped at 12.5% of premium), maker 0.01%; no settlement fee. Deribit: 0.03% maker and taker plus a delivery fee at expiry.
- Access: no KYC by default; the US, Australia (tax residents), Ontario and sanctioned persons are excluded.
Sources: Derive API public/get_all_currencies, public/get_instruments, public/get_vault_statistics, 26.09.2026 · docs.derive.xyz "Supported Products", "Trading Fees", "Borrow Markets", "Common Parameters".
Derive vs Deribit — one table
| Derive | Deribit | |
|---|---|---|
| Who is responsible | Lyra Technologies Corp (Panama); on-chain rules + DAO | Deribit FZE (VARA licence); owned by Coinbase since 14.08.2025 |
| Where the money sits | in a contract on its own chain; withdrawal via bridge in 2–10 min, or 3.5 + 7 days via Ethereum | on the exchange (exchange custody) |
| Option mark | Block Scholes SVI curve, Black formula | own surface from Deribit's book |
| Expiry price | 30-min Block Scholes spot average, 08:00 UTC; 2-of-2 signatures | 30-min Deribit index average, 07:30–08:00 UTC |
| Payout | USDC, straight to the account | BTC/ETH in coin; alts in USDC via a future |
| Margin | standard / portfolio; long options carry no requirement | standard / portfolio; long options carry no requirement |
| Liquidation | auction from 2% to 20% discount over 100 s, then 4h; liquidation fee 2% or 10% — two docs pages contradict each other | staged, run by the exchange |
| Loss beyond the fund | V2: USDC withdrawal fee; V3: haircut on all accounts in the universe | rules allow socialized losses |
| Yield on collateral | USDC 2.05%; yield-bearing tokens accrue in price | USDC 3.40% · USDe 4.75% · stETH 2.25% |
| Alt order books | empty — RFQ only | live orders, 2–8% spreads |
| Incidents | no hacks found 2021–2026; 28h data outage 02–03.07.2026 | daily proof of reserves removed from 01.09.2026 |
| What's ahead | V3 migration with frozen trading and bridges | — |
What would break it, and by how much
- A large borrower defaults in a small universe. After V3 the Alt fund is ≈ $0.35m, RWA ≈ $0.18m. Debt beyond the fund that the auction fails to clear gets haircut from every account in that universe. Check: once the V3 date is announced, pull fund balance and borrowing per universe; the rating for Alt/RWA drops to "no" if the fund is < 3% of borrowing in its universe.
- A botched migration. Trading halted, bridges frozen, withdrawals taking hours to days, some a week. An option expiring during the migration settles late. Check: migration date vs expiry dates of open series.
- USDC pool utilization above 90% — withdrawals wait for loans to be repaid (next check 30.10.2026).
Forecast. On 31.10.2026 the idle-USDC rate on Derive (public/get_interest_rate_history, latest point) will be below the USDC rate on Deribit (public/get_currencies, apr field). Why: today it's 2.05% vs 3.40%, and under Derive's rate formula at up to 80% utilization a supplier earns at most ≈ 3.1% (4.8% × 0.8 × 0.8). If this fails, "Derive adds no yield on cash" rested on one day's snapshot rather than on the mechanics.
What we don't know
No instrument for it: who audited V2 and V3 (V1: Sherlock, Iosiro, Trust Security; no V2/V3 reports found); who holds the two Block Scholes price keys; the distribution and concentration of the $66m in borrowing; actual RFQ prices on series without orders; vault inception dates and therefore annualized yields; whether a standard-margin long option counts toward "account value" in a V3 loss haircut.
Nobody knows yet: the V3 migration date and vote result; what happens to weETH, LBTC and rsETH as collateral.
The docs already describe V3 (testnet) while trading runs on V2. Mark and settlement formulas are identical in both (30 min, Block Scholes SVI), but margin parameters and the loss-sharing waterfall are not.
Separately: Derive's self-reported stats ($32.5m daily premiums, $11.6m open interest) don't reconcile with DefiLlama (≈ $2.2m daily premiums) and are not used here.
Rating and what revokes it
Rating: usable, with limits. Derive makes sense for series Deribit doesn't list; on standard margin with no borrowing; withdrawing the payout on expiry day and not holding open positions through the V3 migration date. DRV staking and yield-bearing collateral don't work as a way to "earn more": Deribit pays more for the same thing.
Date trigger — 31.10.2026, or earlier as soon as the V3 migration date is announced. Revoked by: a migration scheduled inside an expiry window · a universe fund < 3% of borrowing · emergency payout mode · pool utilization > 90% for more than a day · a hack · a change of price signers.
Research for information only. Not investment advice.