Hello,
I’ve been thinking about why crypto needs options when perps already work so well. The answer I keep coming back to is time. You might think bitcoin will be higher in three months, but a leveraged perp can get liquidated tomorrow. Buying a call gives more breathing room for your thesis to play out. You can still lose everything you paid, but a bad afternoon won’t force you out of the trade.
That got us looking at why options are still so small on-chain. We tried buying some, checked the liquidity and looked at why so many earlier attempts failed. Much of it comes back to whether selling options is worth the trouble. Perp exchanges could change that by giving sellers somewhere to hedge and access to traders who are already there. This piece is about why that might work, what still needs fixing and who stands to benefit. We also explore a couple of designs that can incentivise sellers.
For the past couple of weeks, I have been playing around with Papertrade. The house takes the other side of every trade, and every dollar you lose mints you PAPER. The token that earns a share of the house’s fees. Compensating losses with tokens made me wonder what this would look like for options, as we have not yet had a breakout success in that category.
A perp lets you choose how far the price can move against you. It never lets you choose how long you are allowed to be wrong. An option does. In July 2025 we argued that options would follow perps (perpetual futures, referred to as perps throughout the article) on-chain, and could grow a hundredfold in three years. Fifteen months later, they are still under 1% of perp DEX volume. In traditional markets, options trade about three times as much as futures.
This is another article in a series on where crypto options stand.
In the process, I quantify what a perp costs a trader who is right but early, explain why the next wave of on-chain options will be built adjacent to perp markets like Hyperliquid, map who is building what, and ask which protocols benefit most once volume shows up. The short answer is whoever owns the hedge. The positions worth owning are Hyperliquid as the landlord of the hedge market, Derive as the engine others build on, and Paradex as the desk for large block trades. Then we sketch two ventures that could bring in the missing seller.
One puts options inside Metric’s LP pools, and the other is a Papertrade for options.
Right Trade Wrong Route
How often does a perp kill a trade that would have worked? We took BTC’s daily closes from January 2018 to October 2026 and treated every day as a possible entry for a long. We kept only the trades that ended above entry 7 to 180 days later, the ones where the trader was right, and checked whether BTC dipped far enough along the way to liquidate a 5x, 10x or 20x perp.
A 10x perp had a 47% chance of being liquidated over 180 days, 29% over 90 days and 18% over 60 days. An option bought for the same window cannot be closed before expiry. Daily closes also miss the intraday wicks, so the real rates are likely higher. In March, in Bringing the World On-Chain, we wrote that directional flow overwhelmingly chose perps. That still holds for trades measured in hours. But the longer it takes a thesis to play out, the more likely a perp either gets stopped out or liquidated.
We replayed 10 October 2025. At 12:00 UTC, we opened three $1,000 BTC longs. The first was a 10x perp, the second a 30-day call bought with the same $1,000, and the third the same 10x perp with a 7-day put struck 5% below entry, paid for out of the $1,000.
The perp position was liquidated at 21:14. The call fell to $181 at the bottom, but within the hour, it was back above $800. The put cost $52 and kept the 10x long alive through the crash. We repeated the test on eight assets using Deribit prices, and the pattern was consistent.
On alts, protection cost $250 to $340 of the $1,000, because their options price in far more volatility.
Everything on the Menu
Traditional finance and crypto are borrowing from each other, in three ways.
First, traditional finance is importing crypto’s instruments. CME now trades crypto futures and options around the clock. Hyperliquid perp prices show up on the Bloomberg Terminal, and Hyperliquid’s policy arm and trade.xyz have asked the SEC for rules on pre-IPO perps. Polymarket launched perps for users outside the US in September.
Second, crypto is importing traditional finance’s assets. On Hyperliquid, you can trade the S&P 500, Nvidia, gold and oil as perps, at any hour of any day. In Everything is a market, we said treasuries, equities, compute, and attention would all trade on-chain. Anyone in the field can see that every important asset will be on-chain. Perps are how all of that is happening. On 23 March, trade[XYZ]’s WTI and Brent oil perps traded $2.3 billion between them in a single day. There is now a perp on the hourly rental rate of an Nvidia H100 GPU, and one on an index of Pokémon cards. CME and ICE are waiting on the CFTC to list compute futures.
As more prices become tradeable, people will want different ways to trade them. Someone might use GPU perps to bet on a compute shortage. An AI startup exposed to the same price might simply want protection against a higher bill next quarter.Buying an option lets the startup pay a known amount for that protection. If rental rates rise above the agreed level, the option pays out. If they fall, the startup benefits from cheaper compute and loses only the premium. The perp also gives whoever sells that option an instrument to hedge. Each new perp market creates room for an options market alongside it.
Third, traditional finance’s instruments now serve crypto’s own assets. In April, open interest in options on BlackRock’s bitcoin ETF climbed to $27.6 billion and overtook open interest in bitcoin options on Deribit. Coinbase, meanwhile, paid $2.9 billion for Deribit.
The same venue could eventually offer spot, perps, options and binaries on everything from bitcoin to stocks and commodities. Whoever gets there first will pull in most of the flow. We think crypto starts ahead, for two reasons.
Listing is a software decision. IBIT launched in January 2024, yet its options only started trading that November, once the exchange, the clearing house and the SEC had each signed off. Hypercall listed options on SpaceX on 3 June 2026, nine days before SpaceX listed on Nasdaq, using Hyperliquid’s pre-IPO perp as the reference price.
The market never closes. Hypercall lets you trade options on Nvidia and SpaceX on a Sunday, because the perps used to hedge them trade on Sundays too. Spreads outside US market hours are 1.5 to 2 times wider. But no traditional venue offers options on Nvidia at those hours at all.
The rise of perps is the best thing that could have happened to on-chain options, because perps are where option sellers hedge. The deeper perp markets get, the cheaper it is to sell options on top of them. Every new perp market is a potential options market.
Why On-chain Options Kept Failing
Dozens of protocols have tried to bring options on-chain since 2019. Nearly all of them have shut down, pivoted or barely trade. We went through each generation in Trading Optionality, so here is the short version. An option is much harder to list than a perp, for three reasons.
Liquidity is broken across hundreds of order books per asset. A BTC perp has one order book. Deribit lists 884 BTC options across 11 expiries, and someone has to quote every one of them, all the time.
The seller cannot hedge cheaply. A seller has to keep trading the underlying to stay neutral, which needs a deep market right next door.
Selling ties up capital. The first protocols made sellers lock up the most they could ever owe. Selling a $110,000 BTC call for $1,800 meant locking $100,000 to earn 1.8% a month. No professional market maker takes that deal.
Each generation fixed one of these and tripped on another. Hegic’s depositors lost most of about $80 million in May 2021, Lyra’s depositors lost 4.9% in the BTC pool and 17.6% in the ETH pool over seven months, and PsyOptions shut its products in June 2024.
So why would a market maker stay away? Before retail shows up, most of the flow hitting a new venue’s quotes comes from traders arbitraging its prices against Deribit. They only trade when the new venue’s quote is stale, so the market maker loses on almost every fill, even when the venue works perfectly. A venue has two options. It can pay its first market makers to absorb those losses, or bring them retail flow that doesn’t exist yet.
How Today’s Venues Are Solving It
The venues that work today share the same basic design. Orders are matched off-chain, so updating quotes costs nothing. Collateral, margin and settlement stay on-chain, so you keep custody of your funds.
But a better design still has to give someone a reason to sell. The strongest reason is a cheap hedge in the same account. Remember the seller who had to lock $100,000 to sell one BTC call for $1,800?
A portfolio-margin engine considers the maximum possible loss across all positions to calculate the collateral required. If the worst plausible move is 15%, the seller posts about $15,000 instead of $100,000. The premium is still $1,800, so the return on locked capital jumps from 1.8% a month to roughly 12%.
This is why we think options will take off where the perp liquidity already exists, and today that is venues like Hyperliquid and Lighter. An options seller on Hyperliquid gets three things no other on-chain venue offers together.
The first is depth. Hyperliquid has the deepest perp books on-chain, and builder codes let any app plug into them.
The second is one account. A trader who sells a put can reuse the collateral to go long on perps, and spot, perps and event contracts all share one risk engine. This is by far the most critical improvement over predecessors. The third is users, and venues that own the liquidity, the users and the distribution tend to keep the value they create.
Hypercall is the first options venue built on that. A market maker who sells you a call on Nvidia can buy Nvidia perps on Hyperliquid in the same account, and that hedge lowers the margin needed on the option. Its founder reckons 10 to 100 times more market makers quote perps than options. Hypercall hands those perp market makers a reason to quote options too.
The other venues have each found a different seller.
Derive, formerly Lyra, pays for its seller. It cleared about $18 billion of options notional in the year to 19 October, with FalconX as its only publicly named market maker. FalconX’s money stays with its custodian, and Derive shows that balance as a token it trades against, so an institution can quote on-chain without moving collateral on-chain.
Paradex borrows its seller. On 15 September, it plugged in Paradigm’s network, which handles about a quarter of Deribit’s block trades. Daily options volume jumped from under $3 million to $58.7 million in a week, but open interest fell from over $250 million to about $96 million by 14 October. Block traders come for a trade and leave.
Rysk turns holders into sellers. You deposit HYPE, BTC or ETH and sell a covered call to whitelisted market makers for an upfront premium. It did $148 million of notional in the 30 days to 14 October, but volume stalled once the buyers of those calls capped how much HYPE risk they would take.
Front-ends on top of these venues hide the strike grid altogether. Dream lets you top up with Apple Pay, pick a direction and a size, and buys the option for you. Tacticalls, built on Derive, offers three-tap bets and copy trading. Simpler packaging can pull users in. It can’t create a seller when the order books underneath are empty.
Listed But Is It Liquid?
September 2026 was the best month on-chain options have ever had. Notional roughly doubled to about $5 billion, premium rose by a third to $85 million, and open interest peaked near $2.5 billion. That was still only about 7% of Deribit’s volume for the month. Most of it went through Derive.
A venue selling cheap, far out-of-the-money contracts can rack up face value with very little money changing hands. Premium, the cash actually paid for options, is the better yardstick, and on premium Derive’s share is even higher at over 90%.
Hypercall’s notional is mostly cheap, far out-of-the-money options on the S&P 500, bought like lottery tickets. Pink Brains found that $1 of premium on Hypercall bought about $1,000 of face value, against about $50 on Derive. Three things likely explain the divergence.
First, volatility. The S&P 500 moves about 16% a year against BTC’s 43%, so its options are much cheaper.
Second, the contracts. Short-dated calls far from the market price cost a fraction of a percent of face value.
Third, rewards. When Hypercall paid out $1,000 of HYPE a day, notional rose 50-fold while premium barely moved, so traders were likely buying the cheapest contracts to collect rewards. Derive is where traders pay up for real option exposure.
Is Derive’s book good enough? On the evening of 19 October, we pulled every BTC and ETH option listed on both Derive and Deribit, leaving out far out-of-the-money contracts and anything expiring within three days. Deribit had both a bid and an ask on all 346 strikes under consideration. Derive had both on 206. Where Derive did quote, its ask was within 0.2% of Deribit’s, and its bid-ask spread was about 3.8% of the option’s price against 2.6% on Deribit.
In Trading Optionality, published in July 2025, we found Derive’s spreads were often two to five times Deribit’s. The gap has significantly reduced since then.
Away from BTC and ETH, it is a different story. Derive lists options on 12 assets, and we pulled every 25 December strike from its public API.
PUMP, LIT, ADA, VVV and CC have 220 strikes between them for the 25 December 2026 expiry, and one offer. When we tried to buy PUMP calls expiring in December, the order took ages to fill. Even where there are quotes, spreads widen quickly as you move away from the current price. HYPE, the most liquid mid-cap on Derive, shows how fast that happens.
We saw the same thing on every venue we checked. Aevo only quotes memecoin options through its OTC desk, and Dream moved its PUMP and LIT trades to a single OTC counterparty. Listing an option costs a venue next to nothing. Selling one is expensive and risky.
Room to Expire
In traditional markets, 88.65 billion option contracts traded in 2025 against 30.64 billion futures contracts. On-chain, options did $4.9 billion of notional in the 30 days to 19 October 2026, against $627 billion on perp DEXs. That is about 0.8%.
Most of us already hold options without calling them that. If your mortgage lets you repay early, you own a call against the bank. Insurance is a put. Employee stock options are calls paid out as salary. Each one answers a specific need, and every one of those needs shows up in crypto today.
We don’t expect on-chain options to reach traditional finance’s ratio, because some of crypto’s appetite for options is already taken care of by various instruments. Prediction markets sell the binary version, and now perp venues do too. On 19 October, Lighter launched binary options on BTC and ETH.
Even so, three pools of demand are left for on-chain options:
Assets traditional finance cannot list quickly. HYPE, pre-IPO companies, GPU rental rates, and equities and oil on a Sunday.
Users traditional finance cannot serve. Anyone outside the US, anyone without a brokerage account, anyone who wants to trade with $1. We pointed at India’s retail options culture back in It’s All Derivative, where weekly options made up 95% of trades on the NSE.
Options inside other products. Downside protection attached to a perp, covered calls for treasuries, liquidation insurance on a loan. On-chain, an option can be built into any app through an API.
New Markets With Old Winners
The easiest way to answer that is to follow the fees from a single option trade.
The options venue earns a small fee once, when the option trades. The venue where the hedge is placed earns fees again every time the market maker adjusts that hedge. This is why options grow perp volume. Deribit has more than $2 billion of futures open interest that exists largely to hedge options on the venue. So we see three positions worth owning.
Landlords like Hyperliquid own the hedge market, the margin account and the users, and HIP-4 binaries already share the same account. Its HIP-4 contracts are built for “bounded options-like instruments”, which means they could cannibalise Hypercall’s market itself. Either way, the hedging volume ends up on Hyperliquid.
Engines that facilitate options trades are an obvious bet. Derive has the deepest book, a risk engine, a professional user base and, with V3, an API others can build on. If wallets, vaults and agents route their options flow through it, Derive becomes the backend for on-chain options. If they don’t, it stays a venue with subsidised liquidity and one public market maker.
Block desks like Paradex own access to Paradigm’s institutional flow. That is valuable for big trades, and the market makers who come for that flow also deepen the book for everybody.
Venues that don’t offer means for sellers to hedge are already losing ground. Aevo’s options business shrank by more than 90% while the category grew fourfold. Isolated options products are unlikely to stand a chance against products that offer composability with other primitives like perps and spot. It’s likely that we see builder codes like adoption for options where one venue houses the liquidity for various frontends.
The market is already pricing in one of these outcomes. Citrini Research’s “Breaking The Wall” portfolio put 10% in DRV, and DRV is up more than 1,000% this year. At about $390 million of market cap on 22 October against $486,000 of net revenue in the 30 days to 20 October, DRV trades at about 66 times annualised revenue.
What would it take for Derive to grow into that? Say on-chain options get to 5% of perp volume. Perps are on track to do about $660 billion in October volume. 5% of that is about $33 billion in notional options volume. If Derive keeps about 80% of notional and its fee of about 1.1 basis points, that comes to about $35 million of annual fees, or about 11 times fees at today’s market cap. So DRV’s price is betting on a roughly sixfold bigger category with Derive holding its share, while Hyperliquid, Lighter and Paradex chase the same flow.
Options in the Pool
Not every seller has to come from an options venue. Metric is a DEX whose active pools work much like Uniswap’s concentrated liquidity. LPs place liquidity in bins above and below the price and earn the spread whenever traders use their bins. The difference is how bin ranges are defined. On Uniswap, an LP’s range is fixed and only trades move the price, so the LP has to keep resetting the range as the market moves.
On Metric, a curator plugs in a price provider, usually an oracle, and the bins are set as percentages around that reference price. As the price moves, bins move with it. Metric did $4.4 billion of volume in the 30 days trailing 21 October on about $12 million of deposits.
What if those pools quoted options? Metric’s docs have Black-Scholes with volatility inputs as one of the models for a price provider. A curator could price a tokenised ETH call off a live volatility feed.
Take one with a $2,600 strike that expires on 30 October. With ETH at about $2,430 and implied volatility near 50%, Black-Scholes prices it at about $18. If ETH rises to $2,500, the same call is worth about $35 and every bin moves up with it. And because Metric pools route through aggregators such as 0x, 1inch, CoW Swap, and PancakeSwap, traders could buy an option token like any other token, using the swap frontends they already use.
The problem is that this isn’t capital-efficient, even though it would distribute instantly.
Each strike and expiry would be its own token and pool. For example, the ETH $2,600 strike with a 30 October expiry pays holders whatever ETH is above $2,600 at expiry. A new set of tokens is created for the next expiry. A market maker on Deribit quotes hundreds of instruments from one margin account. On Derive, selling the $2,600 call needs about $330 of margin under standard rules, before any offset from hedges. Selling one fully backed $2,600 call locks 1 ETH, about $2,430, to earn $18. Each Metric pool also needs its own inventory, so five strikes across two expiries means ten separate pots of capital.
So the fix is in who issues the token. Say Derive did. Derive already lets sellers post margin and hedge on perps, so if you sold that $2,600 call through it, you would put up about $330 instead of a whole ETH. Metric would then do what it already does well, pricing the token and putting it in front of traders.
Papertrade for Options
Let’s go back to Papertrade. Token incentives to lose pulled in capital and traders fast. That is what on-chain options are missing. Options lack incentives for sellers.
So what would a Papertrade for options look like? A house vault sells calls and puts at listed strikes. When an option expires worthless, its buyer mints the house token in proportion to the premium they lost, and staked tokens earn the house’s surplus.
Option buyers tend to lose frequently. From January 2018 to October 2026, a buyer of a weekly BTC call struck 10% above the market lost their premium in 87% of weeks. Papertrade itself is the obvious candidate to build it out. The difference is that a perp trader with higher leverage is usually liquidated by a small move, while an option buyer stays in until expiry, so the house has to charge a fair premium instead of relying only on liquidations.
The catch is that a call-selling house wins small for months and then loses big when the market moves a lot in either direction. I modelled a cold start on BTC history. A pool starts from zero and sells weekly calls struck 10% above the market. In this simulation, the house collected an average premium of about $1,300 for every $100,000 of BTC, set 50% above what the calls cost it on average. In 12% of weeks, or 57 out of 458, BTC finished above the strike, and the house paid out about $6,900 on average. So net net, the house collected about $1,300 per call, paid out about $860 per call on average and pocketed about $440.
Over eight years, the house made money. But it owed winners 30 weeks’ worth of premium income in March 2021. It couldn’t clear the queue until April 2022. Its worst single week paid out about 20 weeks’ worth of premium income in one go.
The design needs to minimise the queue. Stakers, who earn the house’s surplus in good weeks, should cover its losses in bad weeks, so winners are paid on time instead of waiting. The house should also hedge on the perps that already exist. As BTC rallies toward the strike, it buys BTC perps on Hyperliquid. So in a week with big movement, the hedge pays for much of what it owes winners.
When I added a daily hedge to the same simulation, the worst week cost about 10 weeks’ worth of premium income instead of 20, and the pool never fell into debt during the 2021 rally that had left winners owed 30 weeks’ worth. Our parlays piece made the same point about prediction markets. Someone has to act as the central liquidity manager.
A second design possibility is that instead of only selling, the house takes the other side of every trade, buying options from traders who want to sell and selling to those who want to buy. When buyers and sellers match, the house is flat and keeps the spread, so it only has to hedge and fund the imbalance.
The catch is the token. If losses on either side mint it, a trader could buy a call on one account and sell the same call on another, knowing one of them will lose. Tokens would have to be minted on each account’s net loss per expiry, with a spread wider than the tokens are worth. This is similar to how one can partially hedge their position on Papertrade today.
There are reasons to think this attempt at on-chain options could last. Sellers have deeper markets to hedge in, better ways to use their capital and more routes to reach buyers. Incentives can help get things started. Whether sellers can earn enough to keep quoting when rewards shrink and markets turn is still an open question.
But the foundations are stronger than they were in previous attempts. If they hold, more traders will be able to put a price on how long they are willing to wait for a view to play out. Crypto has made leverage easy to buy. Time could be next.
Still waiting to be right,
Saurabh Deshpande