TL;DR: DVOL is Deribit's 30-day implied volatility index for Bitcoin and Ether, built from options prices. It tells you how large the options market expects moves to be, not which way, so it works best as a regime filter on top of a directional trigger. In Trigr you can add a DVOL node (rising or falling over a lookback you choose) to any strategy, including one that trades a different market, and measure its effect as a labelled experiment.
What is DVOL and what does it measure?
DVOL is a volatility index that Deribit, the largest crypto options exchange, publishes for BTC and ETH. It aggregates the prices of options across strikes to estimate the volatility the market expects over the next 30 days, expressed as an annualized percentage. Deribit's DVOL announcement describes the methodology, which follows the same family of model-free calculations as the Cboe VIX.
Two ideas matter for trading:
- It is implied, not realized. Realized volatility is what price actually did. Implied volatility is what options buyers and sellers are paying for. The two often diverge, and the gap is itself information.
- It is annualized. A reading of 60 does not mean a 60% move is expected tomorrow. It is a yearly standard deviation that you scale down to your holding period.
Converting DVOL into an expected move
Because crypto trades every day, divide by the square root of 365 for a one-standard-deviation daily move, or by the square root of 52 for a weekly move. The figures below are straightforward arithmetic, not forecasts.
| DVOL | Implied 1-sigma daily move | Implied 1-sigma weekly move |
|---|---|---|
| 40 | about 2.1% | about 5.5% |
| 60 | about 3.1% | about 8.3% |
| 80 | about 4.2% | about 11.1% |
| 100 | about 5.2% | about 13.9% |
That translation is useful by itself. If your strategy uses a 2% stop-loss and DVOL implies a 4% daily move, the stop sits well inside ordinary noise. Sizing stops relative to expected volatility is one of the simplest uses of the index.
Why use implied volatility as a filter rather than a signal?
DVOL has no direction. A spike can come from fear of a crash or from demand for upside calls in a squeeze. That makes it a weak trigger on its own and a much more natural filter: a condition that must also be true before a directional entry is allowed.
The reasoning behind most volatility filters is about regime:
- Calm, compressing volatility often accompanies steady trends. Trend-following entries may behave better when implied vol is falling.
- Rising implied volatility signals that options traders are paying up for protection or for large moves. Some traders use it to veto new longs, reduce size or avoid mean-reversion trades that assume a range.
- Implied above realized has historically been the common state in many markets, a gap often called the volatility risk premium. When implied volatility runs far above recent realized moves, the market is pricing stress that has not shown up in price yet.
None of these is a law. Each is a hypothesis about when a given entry rule works better, and the only honest way to use one is to measure it.
Which DVOL strategy ideas are worth testing?
The table lists illustrative structures, not validated edges. Parameters are left open on purpose.
| Idea | Directional trigger | DVOL condition | Rationale | What could break it |
|---|---|---|---|---|
| Calm-trend follower | BTC close above a slow moving average | BTC DVOL falling over 7 days | Trends persist when fear is fading | Vol can fall into a top |
| Stress veto for longs | Any long entry rule | Skip when DVOL rising over 24 hours | Avoid buying into a shock | Fast V-shaped recoveries get missed |
| Breakout confirmation | Price breaks a multi-day high | DVOL rising over 4 hours | Expanding expected moves support breakouts | Vol spikes on downside breaks too |
| Range reversion gate | RSI or Bollinger band extreme | DVOL falling over 30 days | Mean reversion assumes a quiet regime | Regime shifts arrive suddenly |
| Alt beta filter | SOL or ETH momentum entry | BTC DVOL falling over 7 days | BTC options stress spills into alts | Alt-specific news ignores BTC vol |
The last row is a cross-asset idea: the market you trade and the market you read differ. That pattern is covered in more depth in cross-asset crypto signals.
How does the DVOL node work in Trigr?
In Trigr's Studio, DVOL is one of the API data sources you can drop into a strategy graph, next to funding rate, open interest and liquidations. The node reads Deribit's DVOL series at hourly resolution and exposes two settings:
- Condition: rising or falling.
- Lookback: 1 hour, 4 hours, 24 hours, 7 days, 30 days or 90 days.
A node configured as "falling over 7 days" is true when DVOL has declined across that window, and false when it has risen. It can be used as a trigger or a filter, although most people will want it as a filter behind a directional trigger, for the reasons above. The graph structure is one trigger, any filters, a signal and a risk node.
BTC and ETH only, but usable anywhere
Deribit only publishes DVOL for BTC and ETH, so the node only has real data for those two assets. Because any Trigr node can read a different market from the one you trade, a SOL, AVAX or HYPE strategy can still use BTC DVOL as a filter by setting the node's asset to BTC. If you point a DVOL node at an asset with no DVOL series, Studio flags it rather than silently inventing values. The strategy builder docs describe cross-asset nodes in full.
Point-in-time handling
Every input in a Trigr backtest uses only what was known at the bar's close, and the trade fills at the next bar's open. For DVOL on a 4H or daily strategy, that means the latest completed hourly value is carried forward to your bar, never a value from later in the same bar. This matters more for an options-derived series than it might seem: DVOL often jumps during the same hour as a sharp price move, so a backtest that peeks even one hour ahead can look far better than anything you could trade. Point-in-time backtesting explains why this is the main way crypto backtests mislead.
TradFi markets use a different volatility source
For TradFi perps listed on Hyperliquid through HIP-3, such as GOLD, SPX or NVDA, DVOL does not apply. Studio instead offers daily ORATS implied-volatility series for those underlyings, including at-the-money implied volatility, IV rank and a term-structure ratio. For index and commodity underlyings these are proxied from ETF options (for example SPY for SPX and GLD for GOLD). TradFi markets are available for backtests and paper agents only in the current beta.
A worked example: SOL trend-following with a BTC DVOL filter
Here is how you might test the "alt beta filter" idea in a no-code strategy builder:
- Trigger: SOL 4H close crosses above its 50-period EMA.
- Filter: SOL 1D close above its 100-period EMA, so you only trade with the higher-timeframe trend.
- Filter: DVOL node with its asset set to BTC, condition falling, lookback 7 days.
- Signal: long.
- Risk: a modest size, low leverage, a stop-loss wider than the implied daily move, and an ATR trailing stop or time stop so every trade has a defined exit.
Then run it twice: once without the DVOL filter and once with it. You can also describe the idea in plain English to the Studio copilot, or to your own AI assistant through Trigr's MCP connection, and let it draft the graph; check the resulting nodes before you trust the result.
How do you test a volatility filter honestly?
A filter always removes trades. Removing trades almost always changes the equity curve, and it is easy to confuse "fewer trades that happened to be bad" with "a filter that knows something." A few habits help.
Compare as experiments inside one strategy
Keep the base strategy and add the DVOL filter as a labelled experiment rather than saving a new strategy for every lookback. Trigr records experiments in the strategy's version history, so you can compare runs side by side and keep a visible count of how many variants you tried. Iterating with experiments walks through that workflow.
Count your trials
Six lookbacks times two conditions is already twelve variants. If you pick the best of twelve, some of its advantage is luck. That is the problem the Deflated Sharpe Ratio of Bailey and López de Prado was designed for; in Trigr, an ML optimization run reports DSR using its trial count, while a plain backtest does not compute it. The selection bias explainer shows why the best of many backtests usually disappoints.
Look at what the filter actually removed
Open the trade log for both runs. If the filter mostly skipped a handful of large losers clustered in one crash, the improvement rests on very few events. If it removed trades evenly across years and the remaining trades are better on average, the effect is more believable.
Include costs
Switch on slippage and funding before comparing. Trading fees are always applied, but slippage and funding are opt-in and off by default, and a filter that reduces trade count can look better gross simply because it pays fewer costs.
What are the common pitfalls with DVOL?
- Volatility often reacts to the move. Implied volatility usually jumps during or after a selloff, not before it. A "rising DVOL" veto can keep you out of the rebound rather than the drop.
- Levels drift across years. A reading that was high in one cycle can be ordinary in another, which is one reason rising or falling over a lookback is often more robust than a fixed level.
- BTC options are not alt options. Using BTC DVOL on an altcoin assumes BTC stress leads alt behavior. That is often true in broad selloffs and often false for coin-specific news.
- Shorter history. DVOL is a younger series than BTC price data, so a DVOL-filtered backtest has fewer regimes behind it than a price-only one.
Backtests are not guarantees, and perpetual futures are leveraged instruments that can lose more than you expect.
Next steps
Pick one directional strategy you already trust, add a BTC DVOL filter as an experiment, and compare the two net of costs. For more alternative inputs, see how macro data such as the VIX and FRED series can gate crypto strategies.