Crypto runs on two reserve assets, BTC and ETH—and both have halved in value more than once since 2020. Everyone who holds them carries that risk somewhere. Lenders carry it as a haircut, the discount applied to collateral before anyone can borrow against it: Aave, crypto’s largest lending market, will advance about 73 cents against a dollar of WBTC1. Treasuries carry it as runway that can shrink by half in a quarter. Stablecoin issuers carry it as backing that must survive a winter. The volatility is priced into everything, because it has to be.
This article measures what happens when the volatility is removed at the token layer itself, rather than managed with large haircuts after. RiskOFF is the defensive half of The Risk Protocol’s split of BTC and ETH. We compared its volatility with the S&P 500’s and with gold’s, day by day, across six and a half years. The result is not close to what crypto usually looks like.
How we measured it
The RiskOFF numbers in this article come from our own daily record of the token’s value—its net token value, or NTV3—for every single day from December 31, 2019 to May 28, 2026. That is 2,341 consecutive days, covering COVID, the 2021 mania, the 2022 unwind, and two full market cycles. The S&P 500 is daily closing prices from FRED, the Federal Reserve’s public data service, over the identical window; gold, used as a second benchmark throughout, is daily futures closes over the same dates4. Volatility, throughout, is the standard measure every risk desk uses: how widely daily returns swing, scaled to a yearly figure so different assets read on one scale4. Nothing is cherry-picked—no chosen windows, no excluded days. Every day, crashes included.
One deposit, two temperaments
We split a deposit of BTC or ETH into two tokens with opposite jobs. RiskON is designed to deliver ~2X leverage, without the recurring funding rate, forced collateral top-ups, or liquidations that leveraged positions normally carry—which is why we compared it to a 2X Perp when we published The Leverage Tax earlier. RiskOFF takes the other side. Every 30 days it sets a floor 5% below the current price and a cap of about 8% above5; we call these 30-day cycles an epoch. Inside that band, RiskOFF simply tracks the asset. Beyond either boundary, all the upside and downside goes to RiskON.
The part that matters for what follows: the floor is not bought from anyone. It is paid for with the upside RiskOFF gives up past its cap, inside the same pot of collateral. In options language this is a collar—downside protection financed by selling away upside—and this one is costless because the two sides exactly offset, reset at the market price every 30 days. There is no counterparty to trust and no recurring fee bleeding the protection away. It is structural.
Compounding that clamp changes the character of the asset entirely. Here are all 78 consecutive 30-day epochs since the start of 2020—BTC’s raw distribution behind, and what RiskOFF turned each of those same epochs into, in front, on one axis.
ETH, wilder to begin with, compresses the same way.
A token built on top of BTC and ETH that trades like the S&P 500
Over six and a half years, RiskOFF BTC was less volatile than the S&P 500: 19.9% annualized against 20.6%.
That window is not a friendly sample. It contains the fastest crash in BTC’s modern history, a nine-fold rally, and a −77% unwind. Through all of it, a token whose only ingredient is BTC held the volatility of the world’s benchmark equity index—while BTC itself ran at 61% and ETH at 82%. And gold, the asset the world holds when it wants calm, ran at 18.7% over the same window. RiskOFF BTC sits between gold and the S&P 500.
RiskOFF ETH is the harder case, and the honest number is 28.6%—about 1.4× the S&P 500, not level with it. The gap is entirely front-loaded: in 2020 and 2021, ETH ran above 100% annualized volatility and even the clamped token carried 40%. From 2022 onward, RiskOFF ETH has run between 16% and 27% every single year. Both tokens, measured across the full period, carry almost exactly one-third of their underlying asset’s volatility.
Calm when it counts
A single full-period number can hide a lot, so here is the same comparison as a moving picture: rolling volatility, meaning the figure is recomputed each day over the trailing 90 days, so you can watch it change through time.
Two things in that chart deserve to be said plainly. First, the honest one: the S&P 500 in a calm year is calmer than RiskOFF—its median rolling reading (the middle one, across all six and a half years of daily readings) is 14% against RiskOFF BTC’s 19%, and RiskOFF only spends about a third of the sample below it. RiskOFF does not beat the index at its own quietest game.
Second, the one that matters for collateral: RiskOFF’s volatility is stable. The S&P 500’s yearly volatility ranged from 12.7% to 34.7% across this sample—nearly a three-fold swing. RiskOFF BTC’s stayed inside 15.6% to 25.5%. When COVID hit in 2020, the S&P 500 ran at 34.7% for the year; RiskOFF BTC ran at 19.1%. In the year of the fastest crash in modern market history, the calmer of the two was the token built on BTC.
| Year | RiskOFF BTC | RiskOFF ETH | BTC | ETH | S&P 500 | Gold |
|---|---|---|---|---|---|---|
| 2020 | 19.1% | 40.4% | 78.3% | 102.7% | 34.7% | 21.4% |
| 2021 | 25.5% | 42.1% | 78.6% | 104.7% | 13.1% | 15.0% |
| 2022 | 23.4% | 26.5% | 64.6% | 87.7% | 24.2% | 15.5% |
| 2023 | 15.6% | 15.8% | 43.2% | 46.3% | 13.1% | 13.2% |
| 2024 | 17.7% | 19.5% | 52.5% | 64.3% | 12.7% | 15.0% |
| 2025 | 17.9% | 19.0% | 41.9% | 74.6% | 18.6% | 20.5% |
| 2026 | 16.8% | 18.1% | 49.7% | 65.9% | 13.6% | 35.3% |
Annualized volatility by calendar year (2026 through May 28). A holder does not get to choose which year it is: what it needs is an asset whose bad year looks like its good year. Crypto’s worst row here is 104.7%. RiskOFF’s is 42.1%, every row since 2022 is under 27%—and even gold’s yearly range in this sample, 13.2% to 35.3%, is wider than RiskOFF BTC’s.
The drawdown ledger
Volatility is the summary statistic. Liquidation engines—the systems that force-sell collateral the moment a loan turns undercollateralized—care about the specifics: the worst day, the worst month, and the deepest drawdown, meaning the fall from a peak to the lowest point that follows it. Here is the full ledger, same window, no exclusions.
| RiskOFF BTC | RiskOFF ETH | BTC | ETH | S&P 500 | Gold | |
|---|---|---|---|---|---|---|
| Worst single day | −6.8% | −13.3% | −37.5% | −43.2% | −12.0% | −11.4% |
| 1-day VaR, 99% | −2.8% | −4.2% | −8.6% | −11.0% | −3.5% | −3.2% |
| Worst 7 days | −9.6% | −10.7% | −45.3% | −51.6% | −18.0% | −12.0% |
| Worst 30 days | −14.1% | −11.2% | −51.7% | −58.4% | −33.0% | −16.0% |
| Deepest drawdown | −32.6% | −33.9% | −76.7% | −79.4% | −33.9% | −20.9% |
1-day VaR at 99%—value at risk—is the daily loss exceeded on only 1% of days. Drawdowns are measured from each running peak, on daily closes. Worst 7-day and 30-day are the deepest rolling windows (5 and 21 trading days for the index).
RiskOFF BTC’s worst day in six and a half years was −6.8%. The S&P 500’s was −12.0%. Gold’s was −11.4%. BTC’s was −37.5%.
Start with the worst-30-days row, because that is the loss a lender, a treasury, or an issuer actually plans around. RiskOFF BTC’s worst month in six and a half years—March 2020 included—was a loss of 14.1%. The S&P 500’s worst month was −33%. Raw BTC’s was −51.7%.
A fair question at this point: if every epoch has a floor near −5%, how did RiskOFF ever lose 6.8% in a day, or 13.3%? It is because the floor pins where an epoch can settle, measured from that epoch’s starting price—it does not pin the path in between. A token that has climbed toward its cap can give those gains back on the way down, so one bad day can cost more than 5% even though the epoch as a whole cannot settle below its floor.
The crash test
Averages can hide what a crisis feels like. So here are the five worst market events since 2020; the table shows the deepest fall inside each window—from the highest point to the lowest that followed.
| Event | BTC | RiskOFF BTC | ETH | RiskOFF ETH | S&P 500 | Gold |
|---|---|---|---|---|---|---|
| COVID crash, Mar 2020 (46 days) | −52.1% | −15.7% | −61.3% | −17.0% | −33.9% | −11.8% |
| May 2021 cascade (80 days) | −49.3% | −16.3% | −57.0% | −16.2% | −4.0% | −7.6% |
| LUNA collapse, May 2022 (60 days) | −52.0% | −13.9% | −66.2% | −9.3% | −14.7% | −4.1% |
| FTX failure, Nov 2022 (60 days) | −26.0% | −9.7% | −33.5% | −10.9% | −7.3% | −2.1% |
| Yen-carry unwind, Aug 2024 (26 days) | −20.3% | −6.6% | −33.5% | −11.2% | −6.8% | −2.5% |
Peak-to-trough inside each window, daily closes. These are the five deepest market events in the sample; none were skipped.
The pattern repeats in every row. The assets RiskOFF is built from lose a quarter to two-thirds of their value. RiskOFF loses between 7% and 17%—usually less than the S&P 500, and in the neighborhood of gold.
Another way to count the same thing: in six and a half years, BTC had 98 days worse than −5% and ETH had 168. RiskOFF BTC had 3—the same count as gold—and RiskOFF ETH had 13. Days worse than −10%: BTC had 11, ETH had 32, RiskOFF BTC had none. For a lender, those are the days a liquidation engine fires. For a treasury, they are the days the runway shrinks. RiskOFF has barely had them.
Underneath all of these tables sits one number. RiskOFF’s daily beta—how much it moves when its underlying asset moves—is 0.18, on both BTC and ETH. When BTC falls 10% in a day, RiskOFF BTC falls about 2%. That is the kind of behavior that could break the loop that makes crypto crashes feed on themselves, where falling collateral forces sales that push the collateral down further. And RiskOFF BTC’s correlation to the S&P 500 is 0.26, lower than BTC’s own 0.39—so it diversifies a portfolio of traditional assets rather than echoing it.
BTC family
ETH family
What the calm costs
None of this is free, and it is worth being precise about what was paid. One dollar held in RiskOFF continuously—every day, both bears included—grew to $1.87 on BTC and $2.97 on ETH, against $2.34 for the S&P 500 and $10.25 and $15.61 for raw BTC and ETH. RiskOFF surrendered most of crypto’s upside. That is not a flaw; it is the trade—and the surrendered upside does not vanish. It accrues to the other token. Run RiskON the way The Leverage Tax study measures it—held through each of the 13 bull markets that a simple trend rule identifies (price above a rising 200-day average), parked in cash between them—and the same $1 grows to $53 on BTC and $97 on ETH. The chart below puts both halves on one canvas: the split does exactly what it promises. One token concentrates the upside; the other concentrates the stability. RiskOFF’s product is its shape—and its shape is the one thing six years of crypto could not bend.
More than a lending story
Collateral for loans is the sharpest version of the need, but it is not the only one. Anyone who wants to stay in crypto without riding its full swings has the same problem in a different shape.
A protocol treasury holding BTC or ETH is, in effect, betting its payroll on the next crash being gentle. Moving into stablecoins protects the runway but gives up every point of upside—and steps out of the assets the protocol believes in. RiskOFF sits between those two choices: a floor under every 30-day epoch, participation in gains up to the cap, and—in this data—$1 growing to $1.87 on BTC and $2.97 on ETH through two full bear markets, at S&P 500 volatility. Part of a treasury’s runway could live there instead of choosing between all-crypto and all-cash.
Stablecoin issuers face the same shape of problem from the other side: they need backing that holds its value through a crypto winter. And the stable option is not as riskless as it looks—USDC, the second-largest stablecoin, traded near 88 cents during the March 2023 banking scare, a worse day than RiskOFF BTC had in six and a half years. A stablecoin also leaves money standing still: $1 parked there is still $1 years later, while $1 in RiskOFF grew to $1.87 on BTC and $2.97 on ETH.
The same logic reaches trading venues. Exchanges and brokers haircut posted margin—the collateral behind a derivatives position—by how violently it moves, exactly as lenders do. Collateral with one-third the volatility should take smaller haircuts, free more borrowing power per dollar posted, and be far less likely to be force-sold into a falling market. Anywhere BTC and ETH serve as margin today, a calmer claim on the same assets could do the same job with less friction.
The common thread across all of these seats is simple. RiskOFF is a way to store value on crypto rails without carrying crypto’s volatility. Lending is just where that value can be priced most precisely, because lending is where the cost of volatility is published, in public, by formula. So take Aave as the worked example.
Reading the lender’s own ruler
On-chain money markets set collateral parameters through a formal, quantitative process. Aave’s framework—built by Chaos Labs and now operated by LlamaRisk—stress-tests each asset in large-scale simulations: synthetic markets populated with borrower and liquidator agents, price paths fitted to the asset’s real volatility behavior, and forced sales routed through real on-chain liquidity. Out of that come two numbers per asset—the loan-to-value ratio, or LTV, which is how much can be borrowed per dollar of collateral, and the liquidation threshold, the level past which a position is forcibly closed—chosen so that the worst 1% of simulated daily losses stays under a hard bound. Strip the machinery away and two asset-level inputs dominate: how violently the collateral moves, and how deeply it trades. Volatility sets the parameters a mature asset can earn; liquidity decides whether it may earn them at all.
Aave’s own asset framework makes the volatility half of that explicit. Its listing matrix grades assets from A+ to D−, and one of the graded columns is normalized daily volatility—volatility expressed per day rather than per year, so every asset reads on the same scale. Place six and a half years of measured data on that ruler:
On the volatility column of the framework’s own listing matrix, RiskOFF BTC grades in the A band—beside the S&P 500, two full bands above the asset it is built from.
Now put the current parameters next to that. Aave lends up to 80.5% against WETH with a liquidation threshold of 83%, and about 73% against WBTC with a threshold of 78%1. Those numbers price 61% and 82% volatility with fat tails—a far higher chance of extreme moves than a calm distribution would suggest. They are the correct haircuts for the raw assets—that is exactly the point. A collateral with one-third the volatility, a −6.8% worst day, and a worst month shallower than the S&P 500’s is a different risk from the asset it is built on. On the risk numbers alone, there is a case for it to sit in a better band.
What stands between here and a higher LTV
We want to be clear about what this article is and is not claiming. It is a study of a volatility profile, not a listing request—we are not asking Aave, or anyone else, to onboard RiskOFF today. A volatility profile is one of several gates to blue-chip collateral status, and it is the one this data speaks to. The others are earned over time:
- Liquidity. Parameters are only as good as the liquidation route behind them. A recent example shows how seriously risk teams take this: in June 2026, LlamaRisk—Aave’s risk provider—set WETH’s LTV to zero on Aave’s Mantle deployment, meaning WETH there could no longer back new borrowing. The problem was not WETH itself but the local market around it: only about 7.7 WETH of swap depth—the amount that can be sold on-chain without moving the price—was available on that network, too thin to liquidate through safely. Depth gates everything, and RiskOFF’s on-chain depth must be built and then proven under stress.
- Track record. The same listing matrix that grades volatility also grades age, holders, and traded volume; its top grades want years of live history. Six years of NTV data make the case; only battle-tested markets close it.
- Plumbing. We already compute NTV every day; what a lender needs is that number delivered as a robust oracle—the live on-chain data feed a lending protocol reads for prices—plus a liquidation path it can simulate. Integration work rather than invention—the unglamorous part of becoming infrastructure.
Those gates will take time, and they should. But they are operational gates, not structural ones—every one of them closes with adoption. The volatility profile is the part no amount of adoption can manufacture, and it is already in the data: six and a half years, 2,341 days, two full cycles—and RiskOFF exhibits S&P 500-like calm with the most volatile major assets, BTC and ETH, as the underlying.
So here is the claim, stated as precisely as we can make it. Once RiskOFF’s liquidity becomes deep enough to liquidate through and its market history is long enough to grade, a lender applying its own framework consistently should logically assign RiskOFF a higher LTV—and smaller haircuts wherever collateral is haircut—than BTC or ETH themselves. That is not special treatment for a new asset; it is the framework doing what it already does, applied to a different risk profile.
Crypto does not have a shortage of collateral. It has a shortage of collateral that behaves. The calm half was built for exactly that seat. Stability has a new home.
1 Aave V3 Ethereum core market parameters as commonly configured at the time of writing (July 2026); live values are on the Aave parameters dashboard and move with governance. Chaos Labs served as an Aave risk provider from November 2022 to April 2026; LlamaRisk is the primary risk provider today. The volatility bands shown later in the article are from the Aave asset risk matrix as reproduced in the Chaos Labs Aave V3 Risk Parameter Methodology (February 2023).
2 The floor holds on any continuous path, and the protection is honored by the structure itself rather than by anyone’s promise. If the underlying falls to a knock-out barrier roughly 52.5% below an epoch’s starting price, RiskON transfers its share of the underlying to RiskOFF and its own value goes to near zero; the epoch terminates, and a new one starts immediately, with both tokens reset. At the barrier itself, what RiskOFF receives is worth exactly its floor—only a discontinuous gap through the barrier can leave it below that level. No epoch in this six-and-a-half-year sample came close—the deepest declines inside an epoch were −42% (BTC, March 2020) and −51% (ETH, June 2022)—but it is the tail a risk framework would model, and it is why battle-testing belongs on the roadmap above.
3 Every figure uses each token’s Net Token Value (NTV) as our own pricing engine computes it—marked daily, re-struck every 30 days, and applied across the full study history—rather than a simplified model. NTV is built exactly as the protocol prices it: RiskON holds half a unit of the underlying plus a long out-of-the-money call and a short out-of-the-money barrier put, struck each epoch so the two form a costless collar, and RiskOFF holds the mirror image—half a unit of the underlying, short that call, long that put—with the options priced from a two-component GJR-GARCH volatility forecast in a Black-76 and barrier-option framework. NTV is the value a lender’s oracle would reference; a listed token’s market price can trade around it. The study is gross of protocol fees—computed before any fees.
4 Crypto trades every day of the year, the S&P 500 about 252 days, so each series is annualized on its own calendar. Sampling all five series on NYSE trading days only and annualizing everything at √252 gives 19.3% for RiskOFF BTC, 26.9% for RiskOFF ETH, 61.5% for BTC, 82.6% for ETH, and 20.6% for the S&P 500—the same ordering, so the headline is not a weekend-counting artifact. Gold is front-month futures daily closes, annualized on the same 252-day calendar as the index.
5 The −5% floor and +8% cap are the representative levels used throughout this article. In practice, the cap is set afresh at the start of each 30-day epoch so that the collar prices to zero: the premium for the upside RiskOFF gives up is exactly equal to the cost of the protection it receives, so the effective band moves a little from epoch to epoch.