Volatility Is a Surface, Not a Number

An option market does not quote one volatility. It quotes prices across strikes and expiries, which can be expressed as an implied-volatility surface after a pricing convention is chosen. That surface is not merely a more detailed version of a headline index. It is the coordinate system in which downside asymmetry, maturity-specific uncertainty, and event variance become visible. A scalar benchmark can preserve a level; it cannot preserve the geometry that produced it.

This research is about what the average height leaves out. The argument is not that indices are useless; a good index is a disciplined summary, and the most widely watched one is built with real care. The argument is that the surface carries information the summary cannot, and that the difference between them is precisely where event risk, hedging error, and most expensive surprises live.

Section 01Implied and realised

Two quantities share the name volatility and answer different questions. Realised volatility measures a historical price path. Implied volatility is the parameter that reproduces an observed option price under a chosen model. It is forward-looking and risk-adjusted, but it is not a direct physical forecast. The difference between option-implied variance and expected realised variance is the variance risk premium, which reflects compensation for bearing volatility and jump risk as well as beliefs about the future path.

The surface in Figure 3 is therefore a representation of option prices, strike by strike and maturity by maturity. A usable surface must satisfy no-arbitrage constraints across both dimensions and behave coherently under interpolation. Forecast accuracy is only one possible evaluation; internal consistency and hedging behaviour are separate requirements.

Section 02Two axes, one shape

The Black–Scholes model, for all its enduring usefulness, assumes a single constant volatility for an underlying. If that assumption held, every option on the same asset would imply the same number, and the surface would be flat. It is not. Plot implied volatility against strike and you find a smile or, in equity markets, a downward skew: out-of-the-money puts imply higher volatilities than out-of-the-money calls, because the market pays up for crash protection. Plot it against maturity and you find a term structure that can slope up, slope down, or kink. The smile is the market quietly pricing the model's own failure: a correction, expressed in volatility units, for everything the constant-volatility assumption ignores.

Jim Gatheral's The Volatility Surface is the standard practitioner treatment of how these two axes combine into one coherent object and how it must be handled to avoid arbitrage. The key idea for a non-specialist is simply that skew and term structure are not noise around a "true" single volatility. They are the signal. The shape is the information.

The smile is the market quietly pricing the model's own failure: a correction, in volatility units, for everything a single number ignores.

On implied volatility

Section 03What the surface encodes

Read carefully, the surface holds several things a scalar cannot. The slope of the skew encodes the asymmetry of feared outcomes. The term structure encodes when uncertainty is expected to resolve. And localised distortions encode event risk: the bulge in implied volatility around a scheduled catalyst such as an earnings date, a central-bank meeting, or a referendum. On the surface, that event appears as a ridge: a band of maturities whose implied volatility is elevated relative to its neighbours because everyone knows roughly when the uncertainty will strike, even if not how. There is also vol-of-vol (the volatility of volatility itself) and convexity, the curvature that makes options behave non-linearly as the underlying moves. None of these survive compression into one figure.

It is worth being precise about the most famous scalar benchmark. The Cboe Volatility Index aggregates a broad strip of out-of-the-money SPX puts and calls and interpolates between eligible expiries to target constant 30-day expected variance. It is not an at-the-money quote and not one literal point on the surface. It is a carefully specified functional of option prices whose single output still cannot reveal which strikes or nearby maturities produced the reading.

30d_variance  ~=  interpolate_maturity(
                      weighted_sum_of_OTM_option_prices(strikes)
                  )

scalar output does not preserve:
    strike-by-strike skew
    the wider term structure
    local event variance
    surface dynamics through time

Section 04Why a number misleads

The practical danger of the single number is that it invites a single hedge, and a single hedge against a multi-dimensional risk is usually mispriced. A book that is flat to a headline volatility index can still be heavily exposed to a steepening skew, a twist in the term structure, or the collapse of an event ridge once the event passes and its premium evaporates overnight. Hedging error (the residual risk left after a hedge that assumed the wrong dynamics) is born here.

A scalar can preserve the level of a surface. It cannot preserve the shape.

The dynamics are subtler than the static surface. Gatheral, Jaisson and Rosenbaum presented evidence that volatility behaves statistically as a rough process, with a low Hurst exponent relative to smoother stochastic-volatility specifications. That result remains part of an active research programme, but its relevance here is precise: calibration to today's surface does not determine how tomorrow's surface will move. Strike dynamics, term-structure dynamics, and volatility-of-volatility are additional model choices. The surface is the primary cross-section; its evolution is a second object requiring separate evidence.

This research is analysis and commentary for general information. It is not investment advice, an offer, or a solicitation, and it contains no price forecasts. Findings are attributed to their sources; interpretation is the author's.