On-chain options failed retail four times. Here's what we built instead.
You know this story: You believe BTC will hit $100K sometime this month. You act on this by taking a position on this via different instruments - be it Perps, Prediction Markets, or Options. At the end of the month there’s a good-news/bad-news situation: BTC did go over $100K sometime during this month. That’s the good news – that, once again, you were right! The bad news is that, once again, you didn’t make any money off of it, due to liquidations or mistiming the end-of-month price. The gnarly volatility of Bitcoin’s price outsmarted you again.
We built a new product for you – SuperVega. It’s the first product that lets you convert your conviction (“BTC hits $100K sometime this month”) to a one-click action with a transparent one-time premium payment and a pre-defined upside you know when you purchase - via one-touch options.
The following article explains how we got here: by examining past failed attempts at offering options-based products to retail users and trying to fix them. Let’s dive in.
On-chain options failed retail four times. Here's what we built instead.
Every cycle there were attempts to bring options on-chain. And every cycle this failed: the TVL chart peaks and quickly dies. By the next cycle the protocol has pivoted to perps, merged into an exchange, or shut down; examples below.
The standard explanation for this is the hardness of jumpstarting liquidity. "Options need market makers, market makers need flow, flow needs liquidity".
We don’t buy this explanation. Liquidity problems are real, but they’re downstream. The actual problem, in my view, is that nobody started from the view that retail already holds. Everyone started from the instrument - the same 1973 Chicago vanilla options - and then tried to educate retail users to understand it.
This is a post-mortem of four serious attempts and what we are building with SuperVega after internalizing all of it. An autopsy is important because these were products built by serious teams which did their best to build a good product. But the failure points were structural, and mostly invisible until someone hit them at scale. A note on the word "failed": several of the protocols below are still operating - some pivoted, some now serve a different user than they started with. What failed four times was the retail experiment. That distinction is the point of this piece.
The graveyard
Attempt 1: options from a pool - Hegic and the AMM era
The first serious wave, starting in 2020, was peer-to-pool: Hegic was the first project to offer it, then Lyra, Dopex, Premia. Depositors provide capital to a pool; the pool passively writes options to anyone who shows up; buyers get one-click access, LPs get premiums. Hegic pulled $100M TVL almost immediately. It felt like the Uniswap moment for options.
It wasn't, and the reason is specific to options. A spot AMM loses a little to arbitrage and earns it back on uninformed flow. If you quote prices based on an oracle, you’re inviting vol traders to a free lunch. When news breaks and vol spikes, the oracle lags. For those few minutes, you’re selling cheap options to people who price risk for a living. They clean you out, and the LPs hold the bag.
The lesson we take from that era: you can't pool passive sellers against informed buyers and call it a market. Every option needs someone on the other side who looked at that exact risk and chose it, at that exact price.
Attempt 2: options as yield - Ribbon and the DOV (DeFi Options Vault) era
Ribbon Finance defined the DeFi options wave. The pitch was simple and elegant: deposit ETH, the vault sells weekly calls, and you collect premiums as "yield." Peak TVL around $300M within a year of launch. But ultimately two structural problems killed it.
- Everyone sold at the same time. Every DOV auctioned weekly options on Fridays, because Friday expiries matched Deribit liquidity. A synchronized, predictable, massive selling event - that professional trading desks learned to front-run. Implied vol got crushed 30-50% between Thursday evening and Friday morning auctions. Depositors were systematically selling the lows. Paradigm wrote a paper on this ("the Friday problem"). A few vaults did move auction days later. But the structural version of the problem - pooled, scheduled, predictable selling that the other side can see coming - and then front run - never went away.
- The users didn't know they were short vol. "Earn 15% on your ETH" recruits a user who thinks they bought a savings account. What they actually bought was a systematic short-volatility position - steady premium in calm markets, catastrophic in crashes. In May 2022 vault depositors lost double-digit percentages in a week. Nobody was lying - the mechanics were disclosed. But the language that converts ("earn yield") and the risk actually held (short volatility) describe two different products, and in May 2022 depositors found out which one they owned.
Ribbon read the writing early and pivoted to Aevo. Which is the honest conclusion: the retail options-as-yield experiment ended with a retreat to serving pros, and then eventually pivoted to perps.
Lesson: the demand was real - but it was demand for yield, not for short-vol risk. The two look identical until a crash separates them. A seller-side product survives only when the seller knowingly prices the risk they're taking.
Attempt 3: options as perpetuals - Squeeth and the funding-rate wall
Opyn's Squeeth was the most intellectually serious attempt of the era - power perpetuals, a continuous instrument tracking price squared, giving you convexity without expiries or strikes. The design was elegant, and it found real users - 18K users, $1.8B in volume.
It died on one number: cost of carry. Holding convexity means someone has to pay for it continuously, and Squeeth's funding ran 65-180% annualized. The product was a machine that converted your conviction into funding payments faster than the market could validate the conviction. Long-term holders got ground to dust; the only viable users were sophisticated vol traders running it as a hedge leg - the exact opposite of the retail user the simplification was for.
Lesson: you can remove strikes and expiries, but you can't remove the price of convexity. Hide the cost inside a funding rate and it doesn't get smaller - it just gets discovered later, by the user least equipped to model it.
Attempt 4: options as options - the orderbook-and-Greeks generation
Lyra (now Derive), Zeta, Aevo itself - the generation that did the honest thing: put real vanilla options on-chain, with real strike selection, real Greeks, real margin. The infrastructure genuinely works. These are good products for people who already trade options. Most of them are still alive today - this is the emptiest grave in the yard, because the products work. They just don't serve retail.
That's the problem. The addressable market was "Deribit users who prefer self-custody" - a real segment, but small. An option chain presents a retail user with strike selection, expiry selection, IV context, and Greeks - decisions they don't have a framework for. Meanwhile Deribit had a decade head start on liquidity, and perps offered the same directional exposure in one tap.
Here’s what changed my mind: retail does trade options, enthusiastically, when the interface matches how they think. Robinhood cleared $1.7T in options notional in 2023. Retail understands "I pay $ X, I can win $ Y, I can't lose more than $ X" perfectly well. What they don't have is opinions about implied volatility surfaces.
Lesson: the on-chain options stack was never the bottleneck. The instrument's retail interface - clarity about the decisions you need to make - was the problem.
The survivors - and what they prove
It's 2026 and on-chain options are, quietly, having their best year ever. As of mid-2026, on-chain options are clearing roughly $1.9B in monthly notional (July 2026) — the highest the category has ever run.
Derive dominates the category - the large majority of on-chain options notional volume - with a professional cross-margined orderbook on its own L2. Real product, real flow, though a meaningful share of it rides on incentive programs. It's winning the pro trade. That was always winnable; it's Deribit's user with self-custody.
The more interesting case is Rysk. It rebuilt the covered-call trade the DOVs botched, and it's the strongest organic grower in the category - roughly tripling monthly notional over the first half of 2026 - the current time of writing. What it fixed is instructive: users set their own strikes instead of trusting a vault algorithm, market makers compete in an RFQ auction instead of a synchronized Friday event, premium lands in the wallet upfront, and - crucially - it's sold as what it is. An income strategy for holders willing to sell upside, not "safe yield." It fixed the problems of the Ribbon failure.
So the market has learned. But currently, all we have are orderbooks for professionals and income products for sellers of options. The entire 2026 renaissance is a sell-side and pro-side story. The retail buyer - the person with a directional view and $100, the largest untapped audience in this market - is still waiting for a good product to serve it.
Lesson from the survivors: on-chain options work when the product is built around one specific user's actual intent. Nobody has done this for the retail buyer yet.
The actual problem
Everything that failed started from the instrument and worked backwards toward the user. Everything that works started from one user's real intent. Which leads to the trillion dollar question: what is the retail buyer's actual intent?
Listen to what they say - on CT, in Telegram groups, on Polymarket: "BTC tags $120K this month." That's it. Three things: an asset, a price level and a time window. They’re answering the question “will the price of this asset touch this level before this date.” Not "where will the price close on expiry day." Not "what's fair vol."
Suppose you’re one of these retail users, holding such a view – “BTC hits $120K this month” – and consider your options (pun intended) using the current instruments on the market:
| Instrument | What you actually trade | How the path view dies |
| :-- | :-- | :-- |
| **Perp long** | Direction, continuously, on margin | Right about the level, liquidated on the wick before it |
| **Vanilla European option** | Price *at expiry* | Tags your level Tuesday, fades by Friday — right, and paid nothing |
| **Options UI generally** | Your view on vol + strike + expiry | Four decisions demanded, one opinion held |
| **Prediction market** (price markets) | The resolution criterion, often a close | Price markets resolve on the close — the touch itself never pays |To be clear, these aren't failed products - perps are the most successful instrument in crypto, and prediction markets have found real product-market fit for event bets. But for this specific statement, every row breaks the same way: the instrument settles on something other than the thing the user has an opinion about.
The view was right. The instrument was wrong. Four generations of protocols optimized the machinery and left this mismatch fully intact.
To be fair, some builders got close. On-chain binaries - Thales, Buffer - put simple up/down bets in front of retail years ago. But look at the settlement: they resolve on an oracle snapshot at expiry (the endpoint problem again, in a simpler wrapper) or compress into seconds-long windows that are reflex games, not views. And the counterparty is a passive pool quoting off a feed - the Hegic problem wearing a different interface. Even Binance tried this gap from the CeFi side - it launched retail-simplified, American-style options in 2020 with itself as the sole writer, took months of criticism over opaque pricing, and quietly replaced the product with standard European vanillas. When the largest exchange in the industry retreats from a product, the gap isn't imaginary. It's hard.
There is a TradFi instrument that settles on exactly this view - the one-touch. Pays out if price touches the barrier at any point before expiry. It has existed for decades on FX exotics desks. It has never been productized for retail, on-chain or off.
What we built
SuperVega is that instrument, productized. Pick an asset, a target price, a date. Pay a premium. If the price touches your target at any point before the date, you win - a payout fixed and shown to you before you click. If it doesn't, you lose the premium and nothing else. "BTC touches $120K by Friday - pay $100, win $950" is the entire product surface.
Here's why this is different, and why it's better - mechanism by mechanism, each one answering a failure above.
No need to monitor: It settles on the touch, because the touch is the trigger. The moment the price tags your target, that's settlement. Spike Tuesday, full retrace by Friday - you still collect, because your statement was about the path and the path delivered. No other venue gives you this. Perps make you survive the whole path with margin; European options make you price a different condition (the endpoint) than the one you made; prediction markets resolve price bets on the close and price the payout by crowd odds, not vol. We pay a vol-priced payout on the exact statement you made. One decision - level and date - instead of four.
Transparency and bounded downside: It's fully paid, so the position can't die before the view does. The premium is the entire position. No margin, no funding, no liquidation price, no Greeks to monitor. Max loss is known at click-time and never changes. This is the direct answer to both Squeeth and the perp: Squeeth's carry cost and the perp's liquidation are both ways the instrument kills you while your view is still alive. A fully-paid touch position cannot be killed. And unlike the DOV users, our buyers are never short anything: worst case is written on the ticket.
Aligning buyer–seller incentives: Both sides of the trade are ones the market has already validated. Look at who takes the other side. Our writers deposit collateral and earn premium upfront for selling upside above a level they choose - which is precisely the fully-collateralized, self-directed income trade that's driving the only organic growth in on-chain options right now. There is no passive pool getting picked off in the middle: every position has a priced counterparty who chose that strike, that risk, that premium - the exact property the AMM era proved you cannot fake. We didn't invent a new risk and went looking for someone to hold it. We matched the two demands this market has already proven - holders who want paid-upfront income on their upside, and buyers who want a clean path trade - and gave them the venue where the two are the same trade. Read the graveyard again: someone holding a risk they never priced is the flaw that shows up in every collapse above.
A few things we're deliberately not doing. We're not going after Derive or Deribit users; those traders are well served. This isn't a prediction market either; crowd-priced event binaries are a different product, and a good one. And nobody gets promised yield here. Buyers pay a premium for defined upside. Writers underwrite it fully collateralized, knowing exactly what they sold.
The on-chain options graveyard isn't evidence that retail doesn't want options. It's evidence that retail was offered them the wrong way four times. Every generation asked retail to adapt to the instrument, rather than building an instrument they actually wanted. The 2026 survivors prove the fix is building around one user's actual intent. We built for the last unserved user in the market - the one with a view, a hundred dollars, and no instrument that settles on what they actually said.
Today, we're launching public access on Starknet, with EVM wallets and social logins integrated as well. If you've ever been right about a level and still lost money on the trade, we built this for you.