ShortVol as a High-Risk, High-Yield Strategy
Before we position ShortVol as high-risk and high-yield, I tested it five ways. I measured what it earns when the holder sells after a 25% fall and compared that with native DeFi yield. I looked for products that sell volatility this simply. I tested balanced strikes and alternatives to 80/95/105/120. The work covers every BTC and ETH week from January 2020 to July 2026, minute by minute. It also covers over 80 volatility products and 55 native DeFi yields.
a year for BTC ShortVol from January 2020 to July 2026, with all the money moved to USDC whenever it falls 25% in a week
The exit stops each week’s loss near 25%, but losing weeks add up. From October 2020 to May 2021 it sold at a loss in 22 of 32 weeks, and the money fell 97.6% from its peak.
how far $10,000 in BTC ShortVol fell with a 50% exit, to $0.89 by November 2024
A looser exit does not prevent ruin. Every exit tighter than 50% earned more.
what staked ETH, lent USDC, savings rates and Treasury bills pay today
The riskier strategies paid up to 58% in their best full years. Lent BTC earns almost nothing.
products that sell LongVol and ShortVol payoffs as a pair of fungible weekly tokens
Thetanuts does sell ShortVol’s exact payoff, but only as one-off Ranger contracts on Base.
of LongVol needed for each $1 of BTC ShortVol. With balanced strikes it takes $1.00 to $1.73.
Balanced strikes do not cut LongVol’s losses. They only shrink the market.
the strikes I would recommend for BTC and ETH: inner strikes at ±5%, outer strikes at ±22% instead of ±20%
ShortVol then goes to zero in 8 of 343 weeks instead of 14 on BTC, and in 18 weeks instead of 23 on ETH.
What ShortVol earns with exit rules
In this study, 100% of the portfolio is in ShortVol. At each Friday 08:00 UTC open, the money goes into ShortVol. The exit rule checks ShortVol every minute of the week. If it falls 25% below what it cost that Friday, all of it is sold to USDC at that minute’s net token value (NTV). The money stays in USDC until the next Friday open. Then all of it goes back into ShortVol. A second test uses a 50% exit.
I added the 25% exit because, without it, one bad week wipes out the whole portfolio, and bad weeks are common. ShortVol goes to zero whenever the price settles more than 20% away from its open. That happened in 14 of the 343 weeks on BTC, about one week in 25, and in 23 weeks on ETH, about one in 15. With no exit, the BTC money was gone in the tenth week, when BTC fell 43% in the week of 6 March 2020. On ETH the money was gone in the second week.
I recommend we give ShortVol and LongVol holders the ability to set a stop loss. A stop loss sells their tokens to USDC as soon as their value falls by a chosen amount below what they paid.
The test covers all 343 weeks from 3 January 2020 to 31 July 2026, on BTC and ETH, with 80/95/105/120 strikes. Each week settles on the average of the 07:00 and 08:00 UTC prices. I price ShortVol with the risk engine’s own seven-day volatility forecast at each Friday open. I re-price ShortVol at every one-minute price from Coin Metrics, keeping the opening volatility for the whole week. No fees are applied anywhere.
The results, 2020 to 2026
Over the whole period, the 25% rule compounded at 153% a year on BTC and 853% on ETH. These figures assume no fees, and every sale fills at that minute’s value. On the way, the money fell 97.6% on BTC and 98.9% on ETH, and no holder would sit through that.
| BTC | ETH | |||||
|---|---|---|---|---|---|---|
| No exit rule | Exit at 25% | Exit at 50% | No exit rule | Exit at 25% | Exit at 50% | |
| Weeks the exit sold | – | 99 | 55 | – | 137 | 97 |
| Winning weeks | 250 | 225 | 242 | 240 | 199 | 223 |
| Median week | +10.0% | +8.2% | +9.6% | +19.0% | +12.8% | +17.0% |
| Average week | +3.9% | +4.8% | +3.8% | +10.5% | +10.4% | +8.8% |
| Worst week | −100% | −40.5% | −61.5% | −100% | −48.5% | −74.8% |
| Compounded yield a year | −100% | +153% | −65.5% | −100% | +853% | −72.6% |
| $10,000 became | $0 by 13 Mar 2020 | $4.4 million | $9.12 | $0 by 17 Jan 2020 | $27.3 billion | $2.03 |
| Deepest fall | −100% | −97.6% | −99.99% | −100% | −98.9% | −99.999% |
| Losing years, of seven | 7 | 2 | 5 | 7 | 1 | 4 |
$10,000 put in on 3 January 2020 and held to 31 July 2026, 2,401 days.
How can the money fall 97.6% when the exit sells at a 25% loss? The exit limits each week’s loss. It does not limit how many weeks lose. The $10,000 started in January 2020 had grown to $74,358 by 16 October 2020. In the next 32 weeks, the exit sold at a loss 22 times, about a quarter each time. Those sales alone would leave 0.1% of the money. The other 10 weeks all won and multiplied it 18 times. On 28 May 2021, $1,785 was left. ETH fared worse: its exit sold at a loss in 24 of 33 weeks, with 8 in a row.
A single sale can also lose more than 25%, because the exit checks once a minute and the price can jump in between. The worst BTC week lost 40.5%, in the week of 13 October 2023. On 16 October, BTC rose from 5% to 11% above its open in four minutes. At 13:30 UTC ShortVol was at 76% of its cost, just above the line. One minute later BTC had jumped 2.3%, and the sale filled at 60%. Even so, the 25% rule never lost everything in one week.
The 50% rule only made the end slower. A loss of half needs a doubling to win back. At 2020–2022 premiums that took three to four calm weeks. In 2021 the rule lost half in 16 of 53 weeks. The money in BTC ShortVol never ended a week above its $10,000 start. It fell 99.99%, to $0.89 in November 2024, and ended at $9.12, a loss of 65.5% a year. The money in ETH ShortVol rose to a peak of $134,993, then fell to $0.039.
What $10,000 became under each exit rule, on a log scale
BTC
ETH
Single years swing hard. Starting each January with $10,000, the BTC 25% rule turned it into about $591,000 in 2022 and lost 61.1% of it in 2023. Pricing explains the gap. In 2022 the forecast averaged 67%, while BTC’s realised volatility was 56%. In 2023 the forecast averaged 47%, against a realised 51%. When the forecast runs above what BTC does, ShortVol wins. When it runs below, ShortVol loses money.
| Starting with $10,000 each January | BTC | ETH | ||||
|---|---|---|---|---|---|---|
| No exit rule | Exit at 25% | Exit at 50% | No exit rule | Exit at 25% | Exit at 50% | |
| 2020 | −100% | +272% | −95.5% | −100% | +247% | −83.8% |
| 2021 | −100% | −64.7% | −98.3% | −100% | −36.5% | −99.93% |
| 2022 | −100% | +5,805% | +2,765% | −100% | +8,219% | +2,389% |
| 2023 | −100% | −61.1% | −95.4% | −100% | +4,865% | +195% |
| 2024 | −100% | +15.1% | −74.9% | −100% | +405% | −95.0% |
| 2025 | −84.7% | +39.6% | −11.8% | −100% | +429% | −97.7% |
| 2026 (to 31 July) | −100% | +812% | +309% | −100% | +1,022% | +1,974% |
Exits of 20% or tighter earned the most
I tested every exit level from 5% to 95%. Beyond 10%, the yield falls almost steadily as the exit loosens, and every level of 50% or looser loses money on BTC and ETH. ShortVol loses fast, and it earns only one premium a week. On BTC, every level from 5% to 20% made more than 250% a year. At 30% the yield fell to 83.5%, and from 35% the rule lost money. The 5% level sold before settlement in 211 of the 343 weeks, more than half of them.
Compounded yield a year at each exit level, 2020 to 2026
BTC
ETH
Where the yield comes from
ShortVol opens below the deposit. In a calm week it settles at the full deposit and earns the gap. In a typical week of 2020–2022, BTC ShortVol opened at 74% to 82% of the deposit, so a calm week paid 22% to 35%. In 2026 a typical calm week pays 9%, because the yearly average forecast fell from 66% in 2020 to 45%. LongVol buyers pay this gap.
What DeFi pays without emissions
Native yield is cash that someone pays: interest, trading fees, funding, trader losses, option and insurance premiums, liquidation gains and Treasury interest. I left out rewards paid in governance tokens or points. Most staking yield is new ETH or SOL from issuance. Yearly figures come from on-chain indexes where they exist. The rest come from protocol and exchange data, market prices, FRED or DefiLlama’s base yield.
| Strategy and who pays | Last 30 days | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 | 2026 | Worst episode |
|---|---|---|---|---|---|---|---|---|---|
| Staking | |||||||||
| Lido stETHEthereumETH issuance, fees and MEV, less Lido’s 10% fee | 2.2% | – | 5.5% | 4.6% | 4.5% | 3.2% | 2.9% | 2.4% | As 3AC sold, and after Celsius froze withdrawals, it hit 6.5% below ETH on 17 June 2022 |
| Rocket Pool rETHEthereumStaking rewards, less Rocket Pool’s commission | 2.1% | – | 3.7% | 4.3% | 3.9% | 2.8% | 2.6% | 2.2% | |
| Coinbase cbETHEthereumStaking rewards, less Coinbase’s fee | 2.3% | – | – | 4.0% | 3.8% | 2.8% | 2.7% | 2.5% | |
| Frax sfrxETHEthereumAll frxETH validator rewards, paid only to stakers | 2.6% | – | – | 6.6% | 5.7% | 3.7% | 3.3% | 2.9% | |
| ether.fi weETH (restaking)EthereumStaking rewards; restaking pays mostly in tokens, excluded | 2.3% | – | – | – | – | 2.8% | 2.7% | 2.4% | Restaking itself has added about 0%; rival rsETH lost $292M to a bridge exploit in April 2026 |
| Jito JitoSOLSolanaSOL inflation (3.6% now) plus MEV tips and fees | 4.7% | – | – | 6.9% | 6.8% | 8.0% | 7.4% | 5.5% | Mostly SOL inflation, which dilutes unstaked SOL |
| Lending stablecoins | |||||||||
| Aave USDCEthereum (v1, v2, then v3)USDC borrower interest | 4.2% | 4.2% | 6.0% | 1.4% | 3.7% | 7.4% | 4.2% | 3.5% | Never fell, even when USDC hit $0.877 in March 2023 |
| Compound USDCEthereum (v2, then v3)USDC borrower interest | 4.9% | 2.7% | 4.9% | 1.2% | 3.6% | 7.6% | 4.6% | 3.5% | Never fell |
| Morpho Steakhouse USDCEthereumBorrower interest in curated Morpho markets | 4.3% | – | – | – | – | 6.3% | 4.7% | 3.6% | |
| Kamino USDCSolanaUSDC borrower interest | 5.3% | – | – | – | 4.5% | 11.2% | 4.8% | 3.8% | Its extreme week was a high: 30.7% a year in mid-March 2024 |
| Savings and Treasury bills | |||||||||
| Maker DSR, then Sky SSREthereum (Maker’s Pot and sDAI, then sUSDS)Stability fees, Treasury bills and lending | 3.7% | 1.4% | 0.0% | 0.1% | 3.0% | 7.8% | 5.7% | 3.7% | Never fell; DAI hit $0.886 in the March 2023 depeg |
| Maple syrupUSDCEthereumOver-collateralised loans to institutions | 5.2% | – | – | – | – | 10.6% | 7.6% | 4.9% | Never fell |
| Frax sfrxUSDEthereumWhichever pays most: bills, DeFi lending or carry trades | 5.0% | – | – | – | – | – | 6.8% | 4.3% | Frax’s older stablecoin, FRAX, hit $0.880 when USDC depegged in March 2023 |
| Curve scrvUSDEthereumcrvUSD borrower interest | 2.7% | – | – | – | – | 18.0% | 5.3% | 3.1% | Since mid-2023, crvUSD’s lowest daily price was $0.985, on 7 March 2024 |
| Ondo USDYEthereumTreasury bills and bank deposits, less fees | 3.7% | – | – | – | 5.2% | 5.1% | 4.2% | 3.6% | |
| US 3-month Treasury billBenchmarkThe US government; tokenized as BUIDL, USYC and USTB | 4.1% | 0.4% | 0.0% | 2.0% | 5.1% | 5.0% | 4.1% | 3.7% | In October 2026 BUIDL paid 3.5% to 3.8% and USYC 3.3% |
| Synthetic dollars and basis | |||||||||
| Ethena sUSDeEthereumFunding from leveraged longs, staking and reserve income | 4.9% | – | – | – | – | 14.3% | 6.2% | 4.1% | On 10 October 2025 USDe hit $0.65 on Binance but only about $0.99 on Curve |
| Pendle PT-sUSDe (fixed rate)EthereumYield-token buyers paying up front for sUSDe’s yield and, in 2024, points | 5.1% | – | – | – | – | 31.2% | 10.1% | 4.4% | Fixed to maturity; yield-token buyers lost 8% to 69% before points in 13 of 14 expired markets |
| Resolv wstUSREthereumAn ETH and BTC basis book with an insurance tranche | Failed | – | – | – | – | 11.9% | 6.7% | 3.0% | 80M unbacked USR minted on 22 March 2026 (UTC); USR hit $0.20 that day and is $0.07 now |
| Elixir sdeUSDEthereumA basis book that also lent to Stream Finance | Failed | – | – | – | – | 4.1% | 6.1% | – | deUSD traded near $0.001 in November 2025, with most reserves lent to Stream; other holders were offered $1 |
| Stream Finance xUSDEthereumAn undisclosed strategy | Failed | – | – | – | – | – | 16.1% | – | Reported a $93M loss on 3 November 2025; by December xUSD was $0.044 |
| ETH cash and carryBinanceFunding from leveraged longs | 4.1% | 27.4% | 37.5% | 0.8% | 8.3% | 13.0% | 4.9% | 2.0% | Funding was negative 34% of the time during 2022 |
| Leverage loops | |||||||||
| wstETH loop at 3xAave3x staking yield less 2x WETH interest | 2.7% | – | – | 7.9% | 6.1% | 4.2% | 3.3% | 2.4% | WETH borrowing rates reached 103% at the September 2022 Merge and 18% in July 2025 |
| wstETH loop at 10xAave10x staking yield less 9x WETH interest | 4.3% | – | – | – | 12.3% | 7.9% | 5.0% | 2.3% | About −18% a year from 19 April to 7 May 2026, after the rsETH exploit pushed WETH borrowing to 5%–8% |
| JitoSOL loop at 5xKamino (Solana)5x JitoSOL yield less 4x SOL interest | −1.3% | – | – | – | 17.2% | 14.0% | 8.0% | −0.5% | SOL borrowing beat JitoSOL’s yield on 224 of 292 days in 2026, with a peak of 27% |
| sUSDe loop at 5xAave5x sUSDe yield less 4x USDT interest at Aave’s rates (5x only since November 2024) | 5.4% | – | – | – | – | 36.4% | 6.9% | 4.2% | The spread hit −9.3 points in late April 2026: −32% a year that week and −7% a year over four weeks |
| syrupUSDC loop at 5xEthereum5x syrupUSDC yield less 4x Aave’s USDC rate, run on Morpho | 6.2% | – | – | – | – | 18.9% | 13.8% | 6.3% | The spread hit −7.6 points from 21 April to 4 May 2026 |
| Instadapp Lite iETH v2EthereumAn automated stETH loop, measured in stETH | 3.1% | – | – | – | 7.0% | 7.3% | 5.0% | 2.9% | |
| Trader counterparty vaults | |||||||||
| Hyperliquid HLPHyperliquidMarket making, liquidations and a fee share | 4.2% | – | – | – | 94.8% | 58.0% | 17.0% | 9.9% | Deepest fall 5.9%, in late 2023; lost about $4M liquidating a whale’s ETH long in March 2025 |
| Jupiter JLP (fees)SolanaTrading and borrow fees; holders also carry a SOL-heavy basket and trader profit and loss | 7.5% | – | – | – | – | 45.3% | 19.1% | 8.9% | From October 2025 to June 2026 it fell 46.0%, while its basket fell 52.9% |
| Lighter LLPEthereum (zero-knowledge rollup)Market making, liquidations and funding; yearly cells are profit over the average pool value | 20.5% | – | – | – | – | – | 9.4% | 8.7% | Lost 5.35% on 10 October 2025, then shrank from about $700M to $97M after points ended |
| GMX GLP (fees only)ArbitrumTrading fees in ETH, plus trader profit and loss | Closed | – | 28.0% | 25.3% | 16.0% | 14.3% | 12.1% | – | Fell 34.3% with fees from December 2021 to June 2022; a $42M exploit in July 2025 was repaid |
| GMX GM BTC/USD (fees)ArbitrumTrading and borrow fees, plus trader profit and loss | 2.8% | – | – | – | 15.0% | 16.0% | 15.9% | 7.9% | Fell 26.3% between October 2025 and February 2026 |
| Gains gUSDCArbitrumA share of trading fees, less trader profits beyond a buffer | −8.7% | – | – | – | – | 14.6% | 11.3% | 6.5% | Fell 12.0% in the week to 25 September 2026 on traders’ open profits, then regained 11.0% |
| AMM liquidity, fees before price losses | |||||||||
| Uniswap v3 USDC/ETH 0.05%EthereumSwap fees over all pool liquidity | 8.1% | – | – | 44.9% | 28.0% | 32.4% | 23.2% | 22.6% | Fees fell short of arbitrage losses in 2022 and 2023 |
| Uniswap v3 WBTC/ETH 0.05%EthereumSwap fees over all pool liquidity | 8.2% | – | – | 20.4% | 8.1% | 14.7% | 16.1% | 11.2% | |
| Curve 3poolEthereumSwap fees between DAI, USDC and USDT | 0.0% | – | – | 0.3% | 0.5% | 0.9% | 0.2% | 0.0% | CRV emissions were 43% to 60% of its headline yield from 2022 to 2025 |
| Selling options | |||||||||
| Ribbon ETH put vaultEthereumPuts 9% to 15% below ETH; yield in dollars | Closed | – | 19.2% | −27.1% | 6.7% | −3.9% | −22.0% | – | Lost 25.6% of its value in the week to 17 June 2022 |
| Ribbon ETH covered callEthereumCalls 11% to 19% above ETH; yield in ETH | Closed | – | 19.6% | −0.1% | −15.0% | 1.7% | −18.1% | – | Fell 36.1% from its peak in ETH, and lost 12.1% in the week to 9 May 2025 |
| Ribbon BTC covered callEthereumCalls 8% to 16% above BTC; yield in BTC | Closed | – | 2.4% | 10.0% | −23.7% | −6.4% | 0.0% | – | Lost 15.6% in BTC terms in the week to 17 March 2023 |
| Friktion SOL put voltSolanaWeekly SOL puts sold for USDC | Closed | – | – | −45.1% | – | – | – | – | Lost 31.2% in the Terra week and 28.6% in the FTX week |
| Opyn Squeeth CrabEthereumSqueeth funding earned by a delta-hedged short | Closed | – | – | −11.8% | 7.6% | 33.9% | – | – | Crab v1 lost 19.9% in six months |
| Hegic Stake & CoverArbitrumPremiums less payouts, paid in USDC to stakers whose HEGIC covers losses | – | – | – | 10.1% | 7.4% | 2.3% | 3.9% | 1.5% | Lost 8.2% of its stake’s market value in its first epoch as FTX failed; Hegic shows its option book about even since then |
| Underwriting and stability pools | |||||||||
| Liquity v1 Stability PoolEthereumLiquidated ETH, bought with LUSD at up to about 10% off | 0.00% | – | 1.96% | 6.71% | 0.17% | 0.73% | 0.04% | 0.10% | All 318 liquidation transactions paid more than the debt, though in ETH |
| Liquity v2 stability poolsEthereum75% of BOLD borrower interest plus 5% liquidation gains | 4.5% | – | – | – | – | – | 6.4% | 5.8% | No loss; relaunched in May 2025 after a contract issue |
| Aave Umbrella, staked aUSDCEthereumUSDC interest plus a DAO premium for covering bad debt; live since June 2025 | 6.0% | – | – | – | – | – | – | 6.1% | No slash yet; after the rsETH exploit, about 80% of the WETH vault queued to leave in two days |
| Nexus Mutual Foundation poolEthereumCover premiums in NXM, less stake burned for claims | – | – | – | – | 2.8% | 17.3% | 17.8% | −20.7% | One depeg claim burned about 15 months of its 2025 premiums |
| Sherlock staking poolEthereumAudit-cover premiums and interest, plus USDC incentives in 2022 and 2023 | Closed | – | – | −48.2% | −44.9% | 13.3% | 14.7% | 4.3% | A $5M Maple write-off in December 2022 and Euler’s claim in March 2023 |
| Re reUSD (senior)EthereumSenior tranche of quota-share reinsurance, at a set spread over SOFR or sUSDe | 6.9% | – | – | – | – | – | 8.7% | 6.7% | No loss yet; the reinsurer and reUSDe take losses first |
| Re reUSDe (second loss)EthereumThe risk-free rate plus 8.5 points, unless losses reach this tranche | 11.9% | – | – | – | – | – | 15.5% | 13.1% | Re says no underwriting year has reached it yet |
| OnRe ONycSolanaBermuda reinsurance, half property catastrophe and half specialty lines, plus collateral income, after fees | 11.0% | – | – | – | – | – | 9.3% | 11.1% | Four small down days in mid-2025, the largest 0.24% |
| BTC | |||||||||
| WBTC lent on AaveEthereum (v2, then v3)Interest from the few BTC borrowers | 0.003% | 0.760% | 0.062% | 0.141% | 0.191% | 0.143% | 0.019% | 0.006% | $2.8B is lent here. Celsius paid up to 6.2% before it failed in 2022 |
| BTC cash and carryBinanceFunding from leveraged longs | 3.5% | 17.2% | 30.6% | 4.2% | 7.9% | 11.9% | 5.1% | 3.0% | Funding was negative 26% of the time during 2026 |
| BTC quarterly basisDeribitThe quarterly futures premium over spot | 5.4% | 8.5% | 14.5% | 1.5% | 5.2% | 11.9% | 6.8% | 3.5% | Below zero on 27% of days in 2022, and −5.0% a year over 30 days after FTX failed |
| Lombard LBTCEthereumCovered calls on part of the backing, piloted since August 2026 | 0.2% | – | – | – | – | – | – | – | Below its 2.5% target so far |
| ShortVol under the 25% exit rule, simulated | |||||||||
| BTC ShortVolArbitrum (simulated)LongVol buyers’ weekly premium | – | 272% | −64.7% | 5,805% | −61.1% | 15.1% | 39.6% | 4,574% | Deepest fall 97.6%, and the worst week lost 40.5% |
| ETH ShortVolArbitrum (simulated)LongVol buyers’ weekly premium | – | 247% | −36.5% | 8,219% | 4,865% | 405% | 429% | 6,600% | Deepest fall 98.9%, and the worst week lost 48.5% |
All figures are percent a year. Part years are annualised, and 2026 runs only to October. Italic cells are part years for other reasons, such as a launch or a closure. Hover over them for the dates. The ShortVol rows use the 25% exit rule from the first section, compounded within each year to 31 July 2026.
What the table says
The large plain yields now pay about 2% to 5%. Staked ETH pays 2.2%, Aave USDC 4.2%, Sky’s savings rate 3.7%, sUSDe 4.9% and the 3-month Treasury bill 4.1%. Most peaked in 2024. That year Aave USDC hit 12.5% over 30 days and sUSDe paid 14.3%. Staked ETH has paid less every year since 2021, when it paid 5.5%.
Higher yields come from a specific risk, and they have not lasted. A 10x stETH loop made 12.3% in 2023 and 2.3% a year in 2026. A 5x sUSDe loop at Aave’s rates would have made about 36% in 2024 and 4.2% a year in 2026. It lost money in the weeks when borrowing costs spiked. Hyperliquid’s HLP vault made 58% in 2024 and 4.2% a year over the last 30 days. More than half of its lifetime profit came from liquidations. Uniswap fee yields of 8% to 45% are before price losses. In the busiest ETH pool, fees covered only about four fifths of arbitrage losses in 2022 and 2023. In mid-2021 about half of the providers in 17 large Uniswap v3 pools did worse than if they had held the two tokens instead.
Option selling is ShortVol’s closest relative and its clearest warning. Ribbon sold weekly options far from the price, at 10-delta, from 2021 to 2025. A typical week paid 8% to 26% a year. Yet over their lives, its ETH put vault and its ETH and BTC call vaults all lost money in the asset they held. The ETH put vault lost 12.5% a year in dollars. The ETH call vault lost 6.5% a year in ETH, and the BTC call vault lost 5.5% a year in BTC. Only 17 to 20 of their more than 200 weeks lost money. But one week in July 2022 cost the ETH call vault more than 40 weeks of premium. Friktion’s SOL put vault advertised about 30% for 2022 and lost 44%. Opyn’s Crab made about 21% a year from July 2022 to November 2024, but it lost 16.4% in less than three weeks during the FTX crash. Option-vault deposits fell from $651 million in March 2022 to $6.3 million. This is ShortVol without an exit rule.
Underwriting has the same shape. Nexus Mutual’s Foundation pool made about 17% a year in NXM in 2024 and 2025. Then a single claim on its Neutrl USD depeg cover, paid on 15 October 2026, took 22.7% of its stake. Sherlock stakers lost 43.5% when the pool wrote off its Maple deposit. Less than three months later they lost 51% on Euler’s claim, and the pool closed 56% down. Since spring 2025, Re’s reUSDe has paid about 14% a year and OnRe’s ONyc about 10%. Neither has had a losing month yet, but both are less than two years old. Stability pools are the exception. Liquity v1 has never lost money on a liquidation, but it earns almost nothing in calm years. Liquity v2 adds borrower interest and has paid about 6% a year.
Native yield is thinnest on BTC. WBTC lent on Aave pays 0.003%, and most other BTC yield is emissions or points. What remains is funding and selling volatility. Funding paid 3.5% a year over the last 30 days, against 30.6% in 2021. Selling volatility is ShortVol’s market.
Every calendar year from 2020 to 2026 as one dot
Where ShortVol sits
ShortVol is in a different class. The two ShortVol rows at the foot of the table use the 25% exit rule. BTC ShortVol’s calendar years ranged from −64.7% to 5,805%. Plain yields pay about 2% to 5% today, and the riskier strategies paid up to 58% in their best full years. ShortVol’s falls are deeper too. They are deeper than any loss in the staking, lending and savings rows, and deeper than the option vaults’ losses in 2022.
Is anything else this simple?
I checked more than 80 products that sell or trade volatility. To match LongVol/ShortVol’s design, a product needs the first five columns in the table. One pot must back two opposite sides, and anyone must be able to take either side. Both sides must be fungible tokens, with a capped and fully backed payoff. The pair must roll every week on its own. The sixth column shows whether a product was live on 20 October 2026.
The payoff already exists. Thetanuts V4’s Ranger has been live on Base since May 2026. It pays its buyer ShortVol’s exact shape and its seller LongVol’s. But each Ranger is a one-off contract from a market maker, so there is no standing pair. Odette, a partner app built on Thetanuts, sells only the side that wins inside the range. On 20 October, all 42 live Ranger orders came from one market maker and were capped at $10,000 each. Their BTC inner zones were 1.2% to 4.9% wide. Ours is 10%. Thetanuts also deployed a weekly ETH condor vault, the closest vault version of ShortVol. Its strikes are not published, and it has had no deposits since it was deployed in December 2025. Hegic has sold a long condor close to ShortVol since July 2022. But Hegic sells it as NFTs backed by its treasury. The positions close early in the zone of maximum loss, and the long side has no cap.
The split exists too. Volmex and Thales minted two opposite tokens from one pot, and SOFA.org and Stochastic Finance still do. But Volmex tracked implied volatility, not price. Thales paid all or nothing, as SOFA still does, and Stochastic’s two legs bet on direction. None of them offers a standing pair of fungible tokens with linear condor payoffs, open to anyone, fully backed and rolled every week.
| Product | What it pays | One pot | Open both ways | Fungible | Fully backed | Weekly roll | Live | Status and deposits |
|---|---|---|---|---|---|---|---|---|
| LongVol / ShortVolArbitrum | A linear BTC or ETH condor split into ShortVol and LongVol tokens | Yes | Yes | Yes | Yes | Yes | No | Specified |
| Thetanuts V4 RangerBase | The same linear condor, one contract at a time | Yes | Partly | No | Yes | Partly | Yes | Live since May 2026$212K (all of Thetanuts) |
| Hegic Herge strategiesArbitrum | Long condor: sell 10%, buy 20% out of the money; straddles and strangles too | No | Yes | No | Partly | Partly | Yes | Live since 2022$9.3M (all of Hegic) |
| SOFA.org RangeboundEthereum and Arbitrum | All or nothing: the pot if BTC never touches either barrier | Yes | Partly | Partly | Yes | Partly | Yes | Live since June 2024$1.2M (peak $30.2M, 2024) |
| Smilee Impermanent GainArbitrum | Capped two-sided long volatility on WBTC, with a vault on the short side | No | Yes | No | Yes | Yes | No | Stopped rolling in Feb 2025$13K (peak $2.2M, 2024) |
| Volmex v1 (ETHV / iETHV)Ethereum, Polygon and Arbitrum | Long and inverse tokens tracking 30-day implied volatility | Yes | Yes | Yes | Yes | No | No | Dormant$132K (peak $1.7M, 2022) |
| Thales Ranged MarketsOptimism, Arbitrum and Base | IN and OUT range tokens; one pays $1, the other nothing | Yes | Yes | Yes | Yes | Partly | No | Discontinued |
| Stochastic FinanceBase | One USDC mints long and short legs of a bounded call or put | Yes | Yes | Partly | Yes | Partly | Yes | Live since Sep 2026$1.9K |
| Toros Short VolatilityArbitrum | A perpetual delta-hedged short BTC call, with no long twin | No | No | Yes | No | No | Yes | Funded since Feb 2026$670K |
A filled dot means yes, a half dot partly and an empty dot no. Deposits are DefiLlama’s total value locked (TVL) on 20 October 2026, with past peaks where they matter. Toros shows the vault’s own net asset value.
Exchanges and prediction markets sell similar bets
Bybit’s Double-Win has been live since October 2023. It is close to an uncapped LongVol. The stake is lost inside a range. Outside it, the stake comes back with a leveraged gain on the move past the edge. Binance’s Range Bound and the shark fins on OKX, Bybit and Bitget paid more if the price stayed in a band. Most of them were withdrawn between 2023 and 2025. On 20 October 2026, Kalshi listed 188 BTC range contracts for a single hourly fixing. Each paid $1 if the price landed in its range. All of these are positions inside an exchange account, and they jump at the edge of a range. None is a token whose payoff moves in a straight line between the strikes.
Most of these products have closed
More than half of the products I checked have closed, gone dormant or been withdrawn. Friktion closed in January 2023. It cited the FTX and Terra collapses, Solana outages and costs that its revenue no longer covered. Opyn shut Squeeth and Crab in November 2024. Smilee’s WBTC and WETH vaults stopped rolling in February 2025, and CVI’s oracle stopped in February 2026. Sales in Dopex’s ETH straddle vault fell from about 2,540 ETH in January 2023 to about 120 ETH in May, and to under 70 ETH a month from July. Polynomial wound down, saying traders had gone where the liquidity was. GammaSwap judged prediction markets a simpler way to get the same exposure. Few of them were hacked. Most ran out of buyers.
That is the lesson for us. ShortVol’s yield is the premium LongVol buyers pay, so the product lives or dies on LongVol demand. That is where these products failed. What still sells is volatility packaged as something people already want: range bets (Thetanuts, Odette), one-click strategies (Hegic) and upfront premiums for writing options (Rysk, with $48 million deposited). Of these, only Rysk trades in size. I would sell LongVol as a weekly breakout bet, next to prediction markets. I would sell ShortVol as the high-yield side that collects what those bets pay.
Why not balance the two sides?
I would keep ShortVol as the high-risk, high-yield side and sell LongVol as a cheap weekly breakout ticket. The alternative is to set the strikes so both sides have an equal chance. That can mean two things, so I built both. Equal odds means each side beats its cost as often as the other. Of the 63 fixed strike sets I tried, BTC 95/98/102/105 and ETH 92.5/98.5/101.5/107.5 came closest to even, picked on this same history. Equal stakes means both tokens open at half the deposit. The inner strikes sit at the opening price, and the outer strikes are solved for each week.
| BTC | ETH | |||||
|---|---|---|---|---|---|---|
| 80/95/105/120 | Equal odds | Equal stakes | 80/95/105/120 | Equal odds | Equal stakes | |
| LongVol’s share of the deposit at the open | 15.1% | 64.5% | 50.0% | 28.5% | 68.3% | 50.0% |
| Weeks won by ShortVol / LongVol | 250 / 93 | 171 / 172 | 215 / 128 | 240 / 103 | 172 / 171 | 217 / 126 |
| Weeks ShortVol settled at zero | 14 | 138 | 49 | 23 | 126 | 52 |
| LongVol weeks at 5x or more | 30 | 0 | 0 | 13 | 0 | 0 |
| LongVol’s best week | 21.9x | 3.2x | 2.0x | 9.6x | 2.5x | 2.0x |
| $100 a week into LongVol | +$6,483 | −$4,105 | −$3,774 | −$1,303 | −$3,838 | −$4,414 |
| LongVol needed for each $1 of ShortVol | $0.21 | $1.73 | $1.00 | $0.39 | $1.97 | $1.00 |
| ShortVol under the 25% exit rule, a year | +153% | +403% | +217% | +853% | +232% | +857% |
| The same, deepest fall | −97.6% | −99.97% | −99.3% | −98.9% | −99.999% | −99.2% |
343 weeks for each underlying, priced on the engine’s forecast, with the 25% exit rule from the first section.
Balanced strikes do not help LongVol. Putting $100 a week into BTC LongVol made $6,483 with today’s strikes, but lost $4,105 with equal odds and $3,774 with equal stakes. On ETH, today’s strikes lost least: $1,303, against $3,838 and $4,414. Per dollar paid, LongVol got back 82 to 88 cents under every design, so the price sets its loss. The strikes change how that loss falls across the weeks.
They break both products. Equal odds swaps the roles. On BTC, LongVol costs 64.5% of the deposit, and its best week paid only 3.2x. ShortVol becomes a long shot that settles at zero four weeks in ten. Equal stakes still lets ShortVol win about two weeks in three, because small weekly moves are more common than the pricing assumes. It also caps both tokens at double, so LongVol loses its jackpot.
They shrink the market. LongVol buyers are the scarce side. Today each dollar of BTC ShortVol needs 21 cents of LongVol, and each dollar of ETH ShortVol needs 39. Balanced strikes need $1 to $1.97. So the same LongVol demand supports 3 to 8 times less ShortVol.
On BTC they make more, at a cost. The balanced designs make 403% and 217% a year, against 153%. But they fall almost 100% on the way. And the gain needs LongVol buyers to put half to two thirds of the deposit into a capped bet that loses 13 to 15 cents per dollar. On ETH, equal odds makes less and equal stakes about the same.
Which strikes should we use?
We picked 80/95/105/120 without a study. So I tested 42 strike sets for each underlying. The inner strikes ranged from ±3% to ±8%, and the outer strikes sat 10% to 30% beyond them. Every strike set is priced on the engine’s forecast, and every ShortVol yield below uses the 25% exit rule. Then I compared the closest candidates in detail. I also rebuilt the history 2,000 times from random 8-week blocks. This shows whether each difference beats chance.
BTC: move to 78/95/105/122
| Strikes | LongVol costs | Weeks LongVol pays | Best LongVol week | Weeks ShortVol hits zero | ShortVol under the 25% exit rule, a year |
|---|---|---|---|---|---|
| 80/95/105/120 (current) | 15.1% | 40% | 21.9x | 14 | +153% |
| 78/95/105/122 (recommended) | 13.4% | 40% | 21.9x | 8 | +147% |
| 77/97/103/123 | 17.4% | 60% | 11.4x | 7 | +243% |
| 82/95/105/118 | 17.2% | 40% | 19.8x | 18 | +207% |
| 80/96/104/120 | 17.6% | 49% | 15.6x | 14 | +200% |
| 75/95/105/125 | 11.4% | 40% | 21.9x | 6 | +56.5% |
LongVol costs: its median share of the deposit at the open. Weeks LongVol pays: the share of weeks it settles above zero. Weeks ShortVol hits zero: how many of the 343 weeks.
Moving the outer strikes to ±22%, 78/95/105/122, cuts the weeks ShortVol goes to zero from 14 to 8. ShortVol makes about the same, 147% a year against 153%. Today’s strikes come out ahead in only 53% of the rebuilt histories, so the gap may be chance. LongVol gets cheaper, 13.4% of the deposit instead of 15.1%. It still pays in 40% of weeks, with the same best week.
Narrower strikes earn more, because LongVol then costs more and ShortVol keeps more of what LongVol pays. Closer outer strikes, 82/95/105/118, make 207% a year, but ShortVol goes to zero in 18 weeks. Inner strikes at ±4%, 80/96/104/120, make 200% with the same 14 weeks at zero. Across all 42 sets, every one that beats today’s on both yield and weeks at zero has inner strikes at ±3% or ±4%.
One of them, 77/97/103/123, goes to zero in only 7 weeks and makes 243% a year. It beats today’s strikes in 94% of the rebuilt histories. But LongVol buyers pay for it. LongVol costs 17.4% of the deposit, pays in 60% of weeks and its best week falls to 11.4x. $100 a week into it made $1,028, about a sixth of what it made with today’s strikes. LongVol buyers are the scarce side, so I would keep the inner strikes at ±5% and LongVol a cheap breakout ticket.
Wider strike sets do worse. Inner strikes at ±8%, 77/92/108/123, make 25% a year. Outer strikes at ±25%, 75/95/105/125, cut the weeks at zero to 6, but ShortVol then makes 56%.
ETH: move to 78/95/105/122
| Strikes | LongVol costs | Weeks LongVol pays | Best LongVol week | Weeks ShortVol hits zero | ShortVol under the 25% exit rule, a year |
|---|---|---|---|---|---|
| 80/95/105/120 (current) | 28.5% | 53% | 9.6x | 23 | +853% |
| 78/95/105/122 (recommended) | 25.8% | 53% | 9.6x | 18 | +694% |
| 77/97/103/123 | 29.3% | 70% | 6.0x | 15 | +713% |
| 78/93/107/122 | 21.5% | 41% | 16.3x | 18 | +426% |
| 70/95/105/130 | 18.2% | 53% | 9.6x | 5 | +141% |
ETH moves more than BTC, so the same strikes make ETH ShortVol riskier. On 80/95/105/120, ETH ShortVol goes to zero in 23 weeks, against 14 on BTC. Moving the outer strikes to ±22%, 78/95/105/122, cuts that to 18. LongVol gets cheaper, 25.8% of the deposit instead of 28.5%. It still pays in 53% of weeks, with the same best week. But on ETH the move costs yield. ShortVol makes 694% a year against 853%, and today’s strikes come out ahead in three of four rebuilt histories. I would still move ETH, for fewer weeks at zero and one set of strikes for BTC and ETH.
Inner strikes at ±7%, 78/93/107/122, would make LongVol a rarer, cheaper ticket that pays in 41% of weeks. But ShortVol then makes only 426% a year. Moving the outer strikes out to 70/95/105/130 cuts the weeks at zero to 5, and ShortVol then makes 141% a year.
77/97/103/123 makes 713% a year on ETH and goes to zero in 15 weeks. But LongVol then costs 29.3% of the deposit and its best week is only 6.0x. $100 a week into it lost $2,446, the worst of these sets for LongVol buyers.
I would not do two things. The first is to centre the strikes 1% above the opening price, as in 81/96/106/121. That made 174% a year against 153%, but it came out ahead in only 61% of the rebuilt histories. Its extra came from BTC rising in most weeks. It would also give ShortVol a bullish lean and copy the +1% shift in Ultra Bull / Ultra Bear. The second is to scale the strikes with volatility. ShortVol lost money in every version of that I tested in September.
One caution. Under the 25% exit rule, every strike set in the two tables fell more than 95% at some point in the 343 weeks. So the yields turn on a handful of weeks. Steadier numbers drive both recommendations: how often ShortVol goes to zero, how often LongVol pays and what the ticket costs.
What I take from this
The yield is LongVol’s premium
A calm week pays ShortVol the gap between its opening price and the deposit, about 9% at 2026 volatility. LongVol buyers pay that gap. So ShortVol earns most when the engine forecasts more volatility than there turns out to be.
Exits protect a week, not a holder
The 25% exit stopped any one week from taking the money to zero. With a 50% exit, the money in BTC ShortVol still fell 99.99%. Holders should be able to set their own stop loss, at 25% or tighter.
High-risk is literal
Even with the 25% exit, losing weeks add up. The money in BTC ShortVol fell 97.6% from its peak, and in ETH ShortVol 98.9%. Neither got back to its peak until 2022.
Keep ShortVol the high-yield side
Balanced strikes leave LongVol worse off. They cap both tokens at double or swap their roles, and they need 3 to 8 times more LongVol for each ShortVol dollar.
Move both SMART Tokens to 78/95/105/122
Keep the inner strikes at ±5% and move the outer strikes to ±22%. ShortVol then goes to zero in 8 of 343 weeks instead of 14 on BTC, and in 18 instead of 23 on ETH.