Skew stickiness in crypto: why delta-hedging BTC options is not trivial

When spot moves, the smile moves with it, but how it moves decides your hedge. In crypto that behaviour switches regime more often than in FX or equity indices, and a desk that assumes one fixed rule hedges with the wrong delta exactly when it matters.

Empirical observations from VolaLabs surface data. Not investment advice or a recommendation to transact.

Ask a trader what happens to the implied vol of a fixed strike when spot rallies 10%, and you get one of two textbook answers.

• Sticky strike. Each strike keeps its implied vol. The smile stays where it is and spot slides along it, so at-the-money vol changes by the skew.
• Sticky delta (or sticky moneyness). The smile travels with spot. At-the-money vol stays put, and every fixed strike re-prices.

With the usual put skew, the two rules disagree on the sign of what happens at a fixed strike, and so on the vega part of the hedge.

Real markets sit somewhere between, and sometimes beyond, the two textbook cases. The skew stickiness ratio puts the behaviour on one axis: how much at-the-money vol moves per unit of spot move, measured in units of the current skew.

Equity index desks typically see R between 1 and 2: sell-offs lift vol, rallies crush it. FX majors sit closer to sticky delta. Crypto does not stay put. In drawdowns BTC trades like an equity index with put skew bid. In sharp rallies the call wing can richen, skew flattens or flips, and spot-vol correlation turns positive. R moves with it.

The hedge that minimises P&L variance is not the Black-Scholes delta. It adds the vega you pick up because implied vol at your strike moves with spot.

At R = 1 the correction vanishes and the Black-Scholes delta is right. Either side of it, the correction has opposite signs.

A few delta points sound small. On a book that is long vega in the wings and rebalanced several times a day, using the wrong regime adds a systematic error that compounds into hedge slippage.

You cannot read R off a single snapshot. It is a statement about how the surface moves, so you need surfaces close together in time and stable enough that the change you measure is the market, not refit noise.

1. Sampling frequency. A five-minute surface gives hundreds of spot moves a day to regress at-the-money vol changes against, instead of one daily pair.
2. Parameter stability. Where parameters jump between fits, at-the-money vol jumps with them and the regression measures the fitter rather than the market.
3. Per expiry, not per surface. Short-dated and long-dated vol rarely share a regime in the sample. The front end is the first to flip.
4. The sign of skew matters. When the call wing richens, the same R implies the opposite correction, so the two are only interpretable together.

01What VolaLabs provides

The Dynamics metrics report skew stickiness by expiry, the at-the-money skew curve and spot-vol beta, all computed from the same calibrated surface on every update, so the regime in force is a measured quantity rather than an assumption.