Implied volatility

A crypto option chain is quoted in two units at once. The order book shows a premium, often in the coin itself, and every screen beside it shows a percentage. The percentage is the one people compare, and it is also the one that carries the most assumptions.

How it works

An option's price depends on the underlying, the strike, the time left, a rate and one input nobody can observe: how much the underlying will move. Fix the other inputs, take the price the market is paying, and search for the volatility that makes the model agree. What comes out is a price translated into a unit that lets a 60,000 strike and a 90,000 strike be put on one axis, not a measurement of anything that happened.

The price being inverted is frequently in bitcoin. Deribit's inverse options are priced in the base currency, BTC or ETH, with a multiplier of one, so a call is the right to buy one coin at the strike; since August 2026 an in-the-money one settles through an inverse future that pays out in that coin. Its linear options are priced and settled in USDC instead. Those are two different premium currencies for what reads as the same contract, and an inversion has to know which one it is looking at before it can solve anything.

The underlying is a forward, not the index. Deribit states that the implied volatility it displays "uses the forward, rather than the index, as the underlying price in the calculation". Its ticker endpoint carries the choice as data: underlying_index is documented as the "name of the underlying future, or index_price", underlying_price as the price used for the implied volatility calculation, and interest_rate as the rate used in it. A vendor recomputing from the same premium against spot, or with another rate, gets another answer.

One option, three volatilities. The same endpoint returns bid_iv, ask_iv and mark_iv — the volatility implied by the best bid, by the best ask and by the exchange's mark price. The first two depend on somebody quoting; the third is the exchange's own number, and the mark price behind it is what Deribit's option margin formulas use.

On Deribit, volatility is also an order type. An inverse option order can be entered in USD or in implied volatility, and the exchange converts either into the equivalent BTC price and rounds it to a valid tick. So the percentage is not only an analytics column; it is an input a trader types, and what actually rests on the book is the coin-denominated price it was rounded to.

The year has 365 trading days. Deribit's own explanation of DVOL, its 30-day volatility index, tells readers to divide the figure by the square root of 365 for an expected daily move, roughly dividing by 20. The market trades every day of the year, and the calendar the number is annualised over says so.

Why it matters here

A trading-day habit misreads the daily move. Divide an implied volatility of 60 by the square root of 365 and the implied daily move is about 3.1%. Divide it by the square root of 252, roughly the trading days in a year of a market that shuts at weekends and on holidays, and it is about 3.8%. Same number, two readings, and the gap is roughly a fifth. The same trap applies to realised volatility: a crypto series annualised on a 252-day year and set beside an implied volatility annualised on 365 compares two different units. Ask any tool which year it uses before comparing its realised and implied columns.

The weekend is inside the time to expiry. Deribit's published pricing formula measures time to expiry in days over 365 and states that "the full amount of time is included in the calculations"; its contracts trade 24/7 and settle daily at 08:00 UTC. A pricing model that strips weekends out of the time to expiry has less time to fit the same premium into, and so solves for a higher volatility. That is a convention difference, not a signal, and it is invisible unless the vendor writes its calendar down.

DVOL is an index, not an option's IV. It is built from the two expiries either side of 30 days and quoted as a single 30-day figure, and Deribit has launched futures on it. Velo carries it as its volatility index beside ATM implied volatility and 25-delta skew. It is the right number for "how expensive are 30-day options overall" and the wrong one to hold against a specific strike's mark_iv.

Inverse and linear chains do not merge. A BTC-priced option and a USDC-priced one at the same strike and expiry are separate books with separate marks. A loader that keys a surface on strike and expiry alone averages them into something that describes neither.

Which cards carry it at all. Tardis.dev stores the bid, ask and mark implied volatility each venue published, unaltered. Laevitas, Amberdata and Kaiko sell fitted surfaces, which are their own models on top of those quotes. SignalPlus draws a smile for free without saying whose fit it is. TradingView's card records no crypto options chain, IV surface or skew at any price, and CoinGlass's card records five aggregate options endpoints and no implied volatility. Pulling a surface, and why two of them never agree, is its own how-to; the options shelf lists every card that touches the asset class.