At its peak on 22 April 2022, Ribbon Finance held about 310 million dollars, according to DefiLlama, in a business built on vaults that sold options automatically every week. On 12 December 2025, about 2.7 million dollars were drained from its legacy vaults, still live more than two years after the governance vote to fold Ribbon into Aevo, and a 19 percent haircut was proposed for depositors. Between those two dates sits nearly everything a reader needs to know about DeFi options.
The thread of this lesson is one question: who takes the other side of an option, and what do they really earn for it? Automated vaults sold that other side to retail depositors as a yield product. The payoff of that side explains why a falling market hurt, and a badly managed oracle upgrade explains the rest.
This lesson follows our guide to synthetic assets and tokenized stocks, which showed how a contract can copy the price of something it does not hold. Options go one step further, because they sell a right about a future price rather than the price itself. Nothing here is investment advice, and the examples use invented prices unless a date and a source are given.
DeFi options in two sentences, then the real subject
A call is the right to buy an asset at a set price, the strike, until or on a set date, the expiry; a put is the right to sell it on the same terms. The buyer pays a premium up front for that right, and the seller keeps the premium and carries the obligation.
The buyer can lose only the premium. The seller earns only the premium and can lose far more. Sellers are paid for accepting that shape, and option vaults turned it into a percentage on a yield page.
Options differ from perpetual futures, which never expire and keep long and short traders balanced through funding payments. Our guide to on-chain perpetuals and derivatives covers that machinery. An option has a date, and its time value decays as that date approaches. A seller collects that decay.
Who runs on-chain options today, and the numbers
Early DeFi tried to price options with pools, in the spirit of the automated market makers used for spot trading. In Lyra’s design of 2021, liquidity providers funded an AMM that traded options with users and hedged its net delta by trading the underlying asset. The venues that lead today mostly match orders off-chain and settle on-chain, closer to a professional exchange. Most readers meet two names, Derive and Aevo, and their fortunes have split.
Derive: from Lyra to a zkVM exchange
Lyra became Derive on 28 August 2024, in a post that opened with “3 years, 2 bear markets, 1 rebrand. Lyra is now Derive.” The bigger change came in early October 2026 with version 3. According to Bankless, writing on 8 October 2026, V3 replaces Derive Chain, the project’s OP Stack rollup, with a zkVM exchange: orders are matched off-chain, zero-knowledge proofs verified on Ethereum handle margin and settlement, and user funds sit in contracts on Ethereum itself. State data is stored on Celestia, and users can force withdrawals through Ethereum if the operator stalls. DefiLlama records the migration milestone on the same date.
The practical changes for a trader are that ETH and BTC can be margined together, borrowing extends beyond USDC to ETH, WBTC and HYPE, and builders can create fee-earning vaults without writing code. Margin means forced closure is possible, and the logic resembles what our guide to DeFi liquidations describes for lending. Bankless also reports that the app remains closed to US persons.
Aevo: Ribbon’s heir, much smaller
Aevo grew out of Ribbon. Proposal RGP-33, posted on 12 July 2023 and approved in a Snapshot vote that closed on 25 July (133 voters, more than 99 percent of the votes cast in favour), moved the products under the Aevo brand and set a 1:1 migration from RBN to AEVO. Its documentation describes a decentralized derivatives exchange on a custom OP Stack Layer 2, with “off-chain order matching and on-chain settlement”, and lists options, perpetuals, an over-the-counter desk and Aevo Strategies, automated vaults “built by the team behind the OG DeFi Options Vaults (DOVs)”.
On 31 October 2026, DefiLlama showed 176.5 million dollars locked on Derive, about 11 percent below its peak of 198.5 million set on 6 October 2026, and 14.35 million on Aevo, against a peak of 100.2 million on 13 March 2024.
The rest of the field, with dated volumes
| Venue (DefiLlama name) | 30-day notional | 30-day premium |
|---|---|---|
| Derive V3 Options | $2,632.9M | $61.24M |
| Derive Options (older version) | $892.1M | $9.23M |
| Paradex Options | $377.1M | $2.55M |
| Rysk V12 | $135.8M | $1.17M |
| Hypercall | $101.0M | $0.18M |
| CallPut | $95.9M | $0.39M |
| Aevo Options | $64.2M | $1.95M |
| Whole sector | $4,329.3M | $77.07M |
Those figures are the DefiLlama options overview for the 30 days to 31 October 2026. Read the two columns differently. Notional is the size of the underlying position; premium is the money that actually changed hands. Across the sector, premium was under 2 percent of notional, so a headline in billions can describe a business of tens of millions.
By our arithmetic, the two Derive entries made up about 81 percent of notional and about 91 percent of premium, while Aevo Options was around 1.5 percent of notional and 2.5 percent of premium. DefiLlama splits Derive in two because of the October migration, so those shares straddle that event. Rysk and Hypercall, both built on Hyperliquid, took about 5.5 percent of notional between them.
All of this stays small next to centralised trading. Coinbase completed its 2.9 billion dollar purchase of Deribit on 14 August 2025, and The Block’s report gave the exchange 1.185 trillion dollars of trading volume for 2024, without splitting it between options, futures, perpetuals and spot. Annualised, the on-chain sector’s 30-day notional comes to about 53 billion dollars. Different years and different products, so the comparison is rough, yet the gap is more than twentyfold.
DeFi options vaults: the two strategies step by step
An automated option vault takes deposits, sells options on them every week and pays the premium back into the smart contract holding reserves of two or more tokens, supplied by users, that a decentralized exchange uses to execute trades automatically at a price set by a formula.">pool. Ribbon’s documentation is the most detailed public description of the design we found, and Aevo Strategies belongs to the same family. A depositor never picks a strike. The depositor buys a rule for picking strikes and accepts whatever it produces.
Covered call: what the vault does every Friday
In Ribbon’s version, the cycle ran on a fixed schedule:
- Deposits sit idle until the vault rolls, every Friday at 08:00 UTC.
- An algorithm sets the strike shortly before the auction, targeting a 10-delta option, one far enough from the current price that it has a low modelled chance of finishing in the money.
- The vault sells calls on its deposits by auction. Ribbon moved these to Paradigm, in blind, all-or-nothing rounds, so that bidders compete on price.
- The premium goes into the vault, and at expiry a call that finishes out of the money simply lapses.
- If the price finishes above the strike, the vault owes the buyer the difference, paid out of the deposited asset.
- The next week the process repeats. Ribbon charged 2 percent a year as a management fee and 10 percent of weekly premiums as a performance fee, taken only in profitable weeks.
The strike rule, the day and the size of each sale were all public. Anyone registered on Paradigm could bid knowing what the vault had to sell and when, while the vault had no choice about selling.
Cash-secured put: the same machine on the other side
A put vault holds stablecoins and sells puts below the market. Ribbon’s put vaults settled in USDC, so no ether changes hands: if the price ends under the strike, the vault pays the difference out of its stablecoins. Selling puts is often described as “getting paid to buy the dip”. In a cash-settled vault that phrase fails, because the depositor pays for the fall and owns nothing that could recover.
Ribbon’s own documentation is direct: “The primary risk for running a put-selling strategy is that the vault may incur a weekly loss“. Its FAQ says of covered calls that depositors “could potentially give up upside”.
A worked example with example prices
Every number in this example is invented. Take ether at an example price of 2,000 dollars. The covered call holds 1 ETH and sells a call with a 2,200 strike for 0.4 percent, or 8 dollars. The put vault reserves 1,800 dollars and sells a put with an 1,800 strike for 0.5 percent, or 9 dollars. Here is the result after one week, in dollars.
| Spot at expiry (example) | Just holding 1 ETH | Covered call | Cash-secured put |
|---|---|---|---|
| 1,500 | -500 | -492 | -291 |
| 1,800 | -200 | -192 | +9 |
| 2,000 | 0 | +8 | +9 |
| 2,200 | +200 | +208 | +9 |
| 2,500 | +500 | +208 | +9 |
The covered call only trims the holder’s loss on the way down and gives up everything above the strike on the way up. The put earns a fixed 9 dollars in four rows out of five, with a breakeven of 1,791 dollars, then loses 16.2 percent of its reserve in the fifth row. If ether went to zero the loss would reach 1,791 dollars.
A weekly 0.4 percent compounded for 52 weeks gives 23.1 percent. That figure is arithmetic, not a vault’s return. It assumes every week is calm, every premium is the same and no strike is ever crossed. A projected yield tells you what the quiet weeks would pay.
Why many vaults lost money in 2022: capped gains, open losses
The common summary is that vaults lost money by selling volatility in a falling market. That is true of put vaults and misleading for covered calls, and we could not find an audited per-vault return for 2022, so the numbers below are market facts plus one labelled example, not vault results.
| Date | Ribbon TVL (DefiLlama) |
|---|---|
| 31 December 2021 | $285.6M |
| 22 April 2022 (peak) | $310.2M |
| 30 June 2022 | $69.6M |
| 31 December 2022 | $70.7M |
| 30 June 2023 | $27.8M |
| 30 November 2025 | $10.9M |
| 31 December 2025 | $6.4M |
Ribbon’s locked value fell 77.6 percent between the April 2022 peak and 30 June 2022. Ether itself, according to DefiLlama price data, went from about 2,987 dollars on 22 April 2022 to about 1,099 dollars on 30 June, a fall of about 63 percent. staking or liquidity, measured at current market prices. It is the main size metric in DeFi.">TVL is measured in dollars, so that drop mixes the falling price of the deposits with withdrawals, and the two cannot be separated from this series. It is not a measure of depositor losses.
Now follow an example covered-call vault through a similar fall, with rounded prices: 1 ETH at 3,000 dollars, ten quiet weeks at 0.4 percent paid in ether, ending at 1,100 dollars. The call expires worthless each week, because the price is falling. The vault ends with about 1.041 ETH worth 1,145 dollars, down 61.8 percent. The plain holder is down 63.3 percent. In ether terms the vault won. In dollars the depositor who read “yield” on the page lost almost as much as someone with no strategy at all.
The damage comes on the rebound. Suppose ether then jumps to 1,500 dollars in a week, through a strike of 1,210. The vault owes 290 dollars per ETH and is left with about 0.84 ETH, worth about 1,259 dollars, while the holder has 1,500. The cap applies at the low point, and the recovery goes to the buyer of the call.
The put vault takes its hit directly. With ether at 3,000 dollars, a strike of 2,700 and a premium of 13.50 dollars, a slide to 2,000 dollars in one week costs 686.50 dollars, or 25.4 percent of the reserve. Next week the vault sells puts at a lower strike on a smaller balance. A rebound to 3,000 dollars would bring it one more premium, and none of the loss.
Both strategies keep nearly all of the downside and sell off the upside for a small fixed sum. Strong markets and falling markets hurt them in different ways. The shape favours them only when prices drift within a range, and 2022 was no such year.
When the vault itself fails: Friktion and the Ribbon exploit
Market risk was only half of it. Friktion, a vault platform on Solana, peaked at 164.1 million dollars of TVL on 19 April 2022, according to DefiLlama. By Cointelegraph’s report, its vaults moved to withdrawal-only mode from 25 January 2023 and the team announced the end of its user platform the next day, citing a tough market for DeFi growth. A vault that runs itself still needs someone to pay for the front end and the monitoring.
Ribbon’s end was worse. According to The Block, an oracle upgrade on 6 December 2025 created an access-control flaw that, in the words of Anton Cheng of Monarch DeFi, “let anyone set prices for new assets”. On 12 December the attacker pushed false expiry prices for proof of stake coin with a protocol that stakes it for you and issues a transferable token (such as Lido stETH or rETH) representing the deposit plus rewards, which you can use or sell while the coins stay staked.">wstETH, AAVE, LINK and WBTC into the shared oracle, took about 2.7 million dollars from the legacy vaults on Ethereum, then split the funds across 15 addresses. Cheng said the underlying Opyn protocol was not compromised, and the team said Aevo’s Layer 2 exchange was not affected. Our advanced guide to DeFi oracles explains why a price feed is the most dangerous dependency in this kind of smart contract: an option settles on one number, and whoever controls that number controls the payout.
The Block put vault losses at about 32 percent of value. The proposed response was a 19 percent reduction of position value at the time of the hack, with the DAO forfeiting about 400,000 dollars of its own vault positions, which brought net losses to about 2.3 million. The claim window ran from 12 December 2025 to 12 June 2026. After that, the DAO planned to liquidate what remained and pay people who had already withdrawn, up to the missing 19 percent. Aevo also said the DAO “never promised or offered insurance on deposits”. We found no report of the final distribution.
Ribbon’s TVL had shrunk to about 10.9 million dollars by 30 November 2025 and the project had moved on to Aevo, yet the contracts were still live and still receiving upgrades. Depositors who had stopped paying attention were exposed to code that someone else changed.
Principal-protected products: what is protected and what is not
If selling options exposes capital, the opposite design keeps capital safe by buying them. Principal-protected products promise the deposit back and offer a share of any rise. The idea comes straight from traditional structured notes.
The recipe: yield pays for the option
A worked example with invented numbers: a saver deposits 1,000 USDC. The product lends it at 4 percent for a year, so it expects 40 USDC of interest, and it spends that 40 on call options. Long-dated calls on a volatile asset are expensive: take an example price of 400 dollars for a one-year call on ETH, with spot and strike at 2,000 dollars. Forty dollars buys calls on 0.1 ETH.
If ether ends flat or lower, the options expire worthless and the saver gets 1,000 USDC back. If ether rises 30 percent to 2,600 dollars, the options pay 0.1 times 600, or 60 dollars, a return of 6 percent on a move of 30 percent. The saver has a fifth of the rise and has paid for it with a year of interest. In the flat case that 40 USDC of interest is the real cost.
Barriers, coupons and counterparties
Some on-chain products go further and sell options to raise the yield, which quietly removes the protection. Cega built vaults around fixed coupon notes. An analysis of Cega by Paul Veradittakit of Pantera Capital, published in July 2022, described notes holding two or three short put positions, with market makers as counterparties. A knock-in barrier sets how far the underlying may fall. The same piece warned that such notes may not repay the principal in full, with the result tied to the worst-performing asset.
Those short puts are the main source of the return, and the barrier sets how far the market must fall before the seller side of the trade shows up. Cega’s TVL peaked at 40 million dollars on 29 December 2022 and was 0.42 million on 31 October 2026, according to DefiLlama.
What “protected” never covers
In June 2011 the SEC and FINRA warned that any guarantee “is only as good as the financial strength of the company that makes that promise”, that some notes guarantee as little as 10 percent, and that upside may be capped. Their investor alert adds that protection usually applies only if the note is held to maturity, and that holders are unsecured creditors if the issuer goes bankrupt, which is what happened to buyers of principal-protected notes issued by Lehman Brothers Holdings. The on-chain version swaps the bank for a set of contracts, and it keeps the same weak points:
- A bug in the vault or in the lending market that holds the principal is not covered by the word “protected”.
- A stablecoin that loses its peg returns 1,000 units worth less than 1,000 dollars.
- A market maker that owes the vault money can default, and the trades may sit off-chain.
- An oracle error can settle an option on a false price, as Ribbon showed.
- Leaving before expiry usually means selling at whatever the position is worth that day.
The law on crypto options differs by reader
None of this is legal advice. We found no ESMA, AMF or Consob document dealing specifically with options on a DeFi protocol, so what follows covers the general rules.
United States: the CFTC has already acted
On 7 September 2023 the CFTC issued orders against Opyn, Inc. ($250,000), ZeroEx, Inc. ($200,000) and Deridex, Inc. ($100,000). It found that Opyn operated a facility for trading swaps without registering as a swap execution facility, and said of the geoblocking the company had put in place that “those steps were not sufficient to actually block U.S. users from accessing the Opyn Protocol”. Derive’s app, by Bankless’s account, is closed to US persons. A VPN changes where a connection appears to come from. It does not change the rules that apply to the person using it.
United Kingdom: a retail ban that stayed
The FCA banned the sale, marketing and distribution to retail consumers of derivatives, options included, that reference unregulated transferable cryptoassets, from 6 January 2021. It later lifted the part covering certain exchange traded notes, with effect from 8 October 2025, and stated that its “ban on retail access to cryptoasset derivatives will remain in place”.
European Union: MiFID II, not MiCA
MiCA, covered in our dedicated guide, does not apply to crypto-assets that qualify as financial instruments. ESMA’s guidelines of 19 March 2025 separate two cases: a crypto-asset used as the underlying of a derivative, which MiCA does not address, and a crypto-asset that is itself a derivative. Both are assessed against Annex I, Section C of MiFID II, so an option on ether is a MiFID II question.
For retail products, ESMA announced on 27 March 2018 a ban on binary options and limits on CFDs, with leverage of 2:1 on cryptocurrencies. Those measures were temporary, and national regulators then took them over. The French AMF made the binary options ban permanent from 2 July 2019 and restricted CFDs from 1 August 2019; France had already banned electronic advertising for these products through the Sapin 2 law of 9 December 2016. Italy’s Consob did the same on the same two dates, through resolutions 20975 and 20976 of 20 June 2019. Those rules target binary options and CFDs, so they do not automatically describe an ordinary call option in a vault.
Tax: treat it as unsettled
Vault premiums, settled options and vault shares raise questions that a simple buy and sell does not. We found no official guidance written for DeFi option vaults. American readers can start with our US crypto tax guide and British readers with our UK crypto tax guide. Keep a record of every deposit, withdrawal and token received, and ask a professional before filing.
Before touching an options vault: eight checks
This list is a reading exercise for advanced users, not a recommendation to use any product.
- Find out which side the vault is on. If it sells calls or puts, its gain is capped and its loss is not.
- Look for realised weekly results, losing weeks included, rather than a projected APY.
- Identify the counterparty, who buys the options, and how the auction or matching works.
- Read how the settlement price is set and who can change that oracle.
- Check the upgrade powers: who can change the contracts, and with what delay.
- Check the exit: when withdrawals open, and what happens to a deposit between two rounds.
- Confirm that the product is legally open to you, under its terms and the rules where you live.
- Start with a test amount you can afford to lose, and check back after you stop using it, since an old contract can still be changed.
The next lesson of the path turns from options to Pendle, which splits a yield-bearing asset into a principal part and a yield part, and prices the yield as something you can trade on its own.
