Why FX Options Don't Price Volatility the Same at Every Strike

One Number That Refuses to Sit Still

Start with a puzzle. In the textbook model for pricing currency options — the Garman-Kohlhagen model, which is a cousin of the famous Black-Scholes formula — volatility is a single fixed input. Feed in one number for a currency pair, turn the crank, and out comes a price for any strike you like.

Markets, however, did not read the textbook. If you take real FX option prices and work backwards to ask "what volatility would make this price fit the model?", you do not get one number. You get a different answer for every strike. Plotted on a chart, those answers often dip near the middle and curl upward at both ends — a shape that earned the nickname "the smile."

That backward-engineered number is called implied volatility: the market's forecast of future price swings, extracted from the option's own price rather than from history. It is not the same thing as realised volatility, which simply measures how much the currency actually moved over some past stretch. Different contracts on the same currency can carry different implied volatilities because each has its own supply and demand.

What a Smile Actually Looks Like

Plot implied volatility on the vertical axis and strike price on the horizontal axis for one expiry, and a pattern appears. Away from the at-the-money strike — the one nearest the current forward price — volatility tends to be higher.

The shape differs by market. Equity index options in the United States typically slope downward, with lower strikes carrying richer implied volatility than higher ones; that lopsided version is usually called a skew or "smirk." Currency options more often rise in both directions, which is why the friendlier name stuck in FX.

Why the difference? Supply and demand, mostly. Downside puts are popular as portfolio insurance, so buyers bid their prices up; upside calls get bought by people worried about a sharp move the other way. The smile is not a mathematical law. It is a portrait of who wants what, and how badly.

A Brief History of a Wrinkle

Before the 1987 equity crash, the smile in US equity options was relatively flat — traders broadly accepted the constant-volatility assumption. The crash shattered that comfort. Afterwards, out-of-the-money puts became permanently more expensive, since extreme downside moves turned out to be far more plausible than a normal distribution suggested.

The underlying lesson is about distributions. The model assumes returns are log-normal; real returns have fatter tails and skew. When the market's risk-neutral distribution departs from that neat bell shape, implied volatilities differ across strikes. The smile is the visible fingerprint of that departure.

There is also a term-structure dimension. Implied volatility varies with expiry too, often because of scheduled events — an earnings date, a harvest forecast, a central bank meeting. Stack strike and maturity together and you get the volatility surface, a three-dimensional landscape that options desks manage in its entirety.

Quoting a Curve With Three Numbers

Here is the practical problem. You cannot list a price for every strike on earth. So the FX market quotes the smile using a handful of benchmark points per maturity, then fills in the rest by interpolation.

The conventions are worth knowing. FX implied volatilities are quoted against delta, not strike. Delta measures how much an option's value shifts when the underlying moves; in a currency option it also comfortably describes how far out-of-the-money a contract sits. A "25-delta" option is roughly at the point where the market assigns about a one-in-four chance of finishing in the money.

The three benchmarks are:

ATM volatility — the level of the smile at the middle, quoted under a convention called ATM-delta-neutral straddle, where a call and a put with the same strike are packaged so the pair has no net directional exposure.

Risk reversal — the difference between the implied volatility of the 25-delta call and the 25-delta put. Write it as call volatility minus put volatility. A positive reading means the call side is richer, hinting at bullish leanings on the currency pair; a negative reading points the other way.

Butterfly — the average of those two 25-delta volatilities, minus the ATM volatility. This isolates how much the wings sit above the middle. A small butterfly suggests the market expects fairly contained moves; a large one suggests meaningful two-way uncertainty.

A Worked Example, Without the Drama

Suppose the 25-delta call trades at 8.5% implied volatility and the 25-delta put at 7.2%. The 25-delta risk reversal is simply 8.5% − 7.2% = 1.3%, a positive number pointing to a modestly bullish tilt.

Now add the ATM volatility, say 7.8%. The butterfly is the average of the two wings, which is 7.85%, minus 7.8% — leaving 0.05%. That is a very flat smile: the wings sit barely above the middle.

Once you have these three quotes, you can solve for the two wing volatilities directly. The 25-delta call volatility equals ATM plus half the risk reversal plus the butterfly; the 25-delta put volatility equals ATM minus half the risk reversal plus the butterfly. With three strike-and-volatility pairs in hand, a curve can be fitted and volatilities read off for any strike.

The Arithmetic Has Rules

Fitting a curve is one thing; fitting a sensible one is another. Prices that permit arbitrage — free money by construction — are a real hazard when you extrapolate into the wings, especially under stress.

Academic work on arbitrage-free smile construction shows a neat route: under certain assumptions about how implied volatility moves in the near term, arbitrage-free FX volatilities emerge as the roots of a cubic polynomial with a symmetry property. That symmetry lets risk reversal and butterfly prices be identified from the same polynomial's roots, while the ATM volatility is matched by construction. The approach works directly with non-premium-adjusted deltas, so forward rates and strikes are not required as inputs.

A widely used alternative, the Vanna Volga approach, builds the smile from three benchmark options and is regarded as very accurate for interpolation. But it is an interpolation and extrapolation method rather than a coherent model, and arbitrage-free prices are not guaranteed. Empirical comparisons between the two approaches find their results broadly similar when interpolating between quoted points, with differences appearing when extrapolating into the wings — and those differences are more pronounced on a stressed trading day than a calm one. There is a further practical catch: when near-term volatility is large, the smile can become too steep to price deep out-of-the-money puts sensibly.

Why the Market Has Been Busy

None of this is abstract bookkeeping. The global FX market averaged more than $7.5 trillion per day in April 2022, a volume roughly 30 times daily global GDP, and turnover has historically risen alongside volatility as participants adjust positions, hedge currency risk and disagree with each other about the future.

Options activity specifically has grown sharply. In the 2025 BIS Triennial Survey, FX options volumes doubled from $303 billion to $634 billion per day, reaching about 7% of total FX turnover. One important caveat: that survey month coincided with a burst of trade-policy announcements and elevated volatility, so some of the jump may reflect event-driven activity rather than a durable trend. It is genuinely hard to separate the two — a point worth carrying into any single month's data. Context also matters: the 2022 survey month was itself unusually volatile, so comparisons between survey years are not like-for-like.

Structural shifts matter too. More trading has migrated toward bilateral and disclosed venues, and inter-dealer activity has risen. Some of this reduces the visible price information available to the wider market, which complicates price discovery and monitoring. Currencies remain concentrated as well — the US dollar sat on one side of roughly 90% of all FX trades in April 2022.

Reading the Shape, Not Predicting the Future

The smile and its two headline numbers are best understood as a summary of market conditions, not a forecast you can bank. Risk reversals and butterflies describe how option prices are distributed across strikes right now — the balance of hedging demand and conviction. They are useful context. They are not signals.

Several honest caveats are worth holding onto. Conventions vary by currency pair, maturity and jurisdiction, and the choice between spot and forward deltas, or simple and premium-adjusted ones, changes the arithmetic. The 25-delta benchmarks are liquid because they sit near the money; deep wings are thinner and their inferred volatilities less reliable. And volatility can be elevated or depressed for entirely justified reasons, so a rich reading does not automatically mean an opportunity.

Perhaps that is the most useful takeaway from the whole apparatus. A smile looks cheerful on a chart. What it is really telling you is where the market is paying up for protection, and where it is not — a picture built from supply, demand and the occasional unpleasant surprise.