How Vol Surfaces Are Built
You don't see a continuous surface in the market. You see scattered quotes at a handful of strikes and expiries. The "surface" is constructed by converting those quotes to implied volatility, filling the gaps, and enforcing consistency rules so no one can extract free money.
This page walks through that pipeline: raw quotes in, smooth surface out.
Step 1: Start With Sparse Quotes
Option markets don't quote every possible strike and expiry. On any given instrument you might see quotes at 15-20 combinations out of hundreds of possible grid points. Most of the surface is empty.
From Quotes to Surface
| Strike | 7d | 14d | 30d | 60d |
|---|---|---|---|---|
| $85k | $120 | $340 | -- | -- |
| $90k | $450 | $820 | $1,400 | -- |
| $95k | $1,200 | -- | $2,800 | $4,200 |
| $100kATM | $3,500 | $4,800 | $6,200 | -- |
| $105k | $1,800 | $3,100 | -- | $5,800 |
| $110k | $650 | -- | $2,100 | -- |
Click through the three steps above. Notice that in Raw Quotes, most cells are blank. In Extract IV, we invert each price through Black-Scholes to get an implied volatility, but the same gaps remain. Only after Interpolate does every cell fill in, and the cells marked "SVI" were not observed in the market.
This is the central challenge: turning sparse, noisy observations into something smooth and internally consistent. Let's walk through how each step works.
Step 2: Convert Prices to IV
Each quoted option price needs to be converted into an implied volatility: the that makes Black-Scholes match the market price.
There is no closed-form solution. We solve numerically: try a volatility, compute the BS price, check if it matches, and narrow the range. Watch the solver converge step by step:
IV Root-Finding: Watch the Solver Converge
The solver narrows a range of possible volatilities until BS price matches the market price.
| Step | Try σ | BS Price | vs Target | Range |
|---|---|---|---|---|
| 1 | 150.50% | $11278 | > too high | [1.0%, 300.0%] |
| 2 | 75.75% | $5825 | > too high | [1.0%, 150.5%] |
| 3 | 38.38% | $2995 | < too low | [1.0%, 75.8%] |
| 4 | 57.06% | $4421 | > too high | [38.4%, 75.8%] |
| 5 | 47.72% | $3711 | > too high | [38.4%, 57.1%] |
| 6 | 43.05% | $3354 | < too low | [38.4%, 47.7%] |
| 7 | 45.38% | $3533 | > too high | [43.0%, 47.7%] |
| 8 | 44.21% | $3444 | < too low | [43.0%, 45.4%] |
| 9 | 44.80% | $3488 | < too low | [44.2%, 45.4%] |
| 10 | 45.09% | $3511 | > too high | [44.8%, 45.4%] |
| 11 | 44.94% | $3499 | = match | [44.8%, 45.1%] |
The yellow line is the market price. Each guess (dot) evaluates BS at that volatility. The shaded region shows the remaining search range narrowing with each step.
Change the inputs to see how the solver behaves with different strikes (OTM vs ATM), maturities, and prices. Notice how the search range (shaded region) halves with each step. Most production systems use Brent's method (guaranteed convergence) or Jäckel's rational approximation (machine precision in one step).
After inverting every quoted price, we have a sparse grid of IV values. But those values are not constant across strikes. Before we can fill the gaps, we need to understand the shapes the IV curve naturally forms.
Step 3: Understand the Shapes
If the world matched Black-Scholes perfectly (lognormal returns, constant volatility), implied vol would be identical at every strike. It is not. The discrepancy tells you something real about the market, and understanding these shapes is essential for choosing the right interpolation.
Why IV varies by strike
Volatility is random. Black-Scholes assumes it is constant. When vol itself fluctuates, OTM options become more valuable than Black-Scholes predicts. An OTM option benefits asymmetrically from vol changes: if vol rises, the option moves closer to the money and gains rapidly; if vol falls, the option is already near-worthless and loses little. This convexity means OTM options are worth more when vol is uncertain, which shows up as elevated wing IV. The more vol fluctuates (higher "vol of vol"), the wider the smile.
Spot and vol move together. The smile alone would be symmetric. The skew comes from the fact that, in most markets, when prices drop, volatility rises. Traders call this "up the escalator, down the chute." OTM puts (which pay off in a crash) are worth more than Black-Scholes thinks, because the crash will come with a vol spike that makes them even more valuable. OTM calls are worth less, because rallies tend to compress vol. The result is the put smirk: the left wing trades at higher IV than the right wing. For more on smile shapes, see the volatility course lesson on smile and smirk.
The four shapes you'll see
Smile and Smirk Shapes
Click through each shape. The 25-delta risk reversal (bottom left) measures the tilt: positive means puts are richer, negative means calls are richer. The butterfly (bottom right) measures curvature, independent of direction.
Skew Visualization
Drag the slider to see how skew changes the IV curve across strikes. Put skew (positive RR) is normal; call skew is rare.
Drag the slider to see how skew strength changes the curve. A risk reversal near zero (flat) means the market has no strong directional bias.
How IV varies by time
The surface has a second dimension: the term structure of volatility.
Term Structure
Backwardation: Near-term IV > far-term. Signals event risk priced in.
Toggle between shapes to see how term structure changes. Backwardation often signals an upcoming event.
| Shape | Meaning | When You See It |
|---|---|---|
| Contango | Long-dated IV > short-dated | Calm markets. Vol expected to mean-revert upward. |
| Flat | Similar IV across maturities | No strong term view. |
| Backwardation | Short-dated IV > long-dated | Near-term event risk. Something specific is driving front-month vol. |
Backwardation is the key signal. When short-dated vol spikes above long-dated vol, the market is pricing a specific near-term catalyst. After the event passes, the term structure typically snaps back to contango. See vol regimes for how the entire surface shifts across market conditions.
The term structure also encodes forward vol: the market's expectation for vol between two future dates. If 30-day IV is 52% and 90-day IV is 48%, the implied forward vol from day 30 to day 90 is:
Much lower than either spot vol. The market is saying: "The next 30 days will be wild, but after that, things calm down." Try computing forward vol with different inputs in the Equation Explorer (select the "Forward Vol" tab). If the forward variance goes negative, that signals a calendar arbitrage.
Now that we understand what shapes the surface should form, we can fill the gaps intelligently.
Step 4: Fill the Gaps
We have scattered IV observations and we know what shapes they should form. The challenge: fill every empty cell while keeping the surface smooth, stable, and arbitrage-free.
Interpolation Methods Compared
The white dots are the only real market observations. Everything between them is estimated. Click each method to see its strengths and weaknesses.
Two rules for interpolation
Nassim Taleb frames it with two principles:
1. Eliminate jaggedness. Linear interpolation creates sharp corners. The market cannot be expected to have discontinuous jumps in implied vol between adjacent strikes. Smooth the corners.
2. Adapt to the market. If the market uses a particular interpolation convention, your pricing engine should match it. Your risk management can use whatever method you think is most accurate, but your quotes need to agree with the market's convention, or you will get adversely selected.
What can go wrong
Arbitrage. If a $97k call is priced too cheap relative to $95k and $100k, a trader can buy the $97k and sell a blend of the other two for risk-free profit (butterfly arbitrage). If the surface implies that total variance decreases with maturity, a calendar arbitrage exists.
Nonsensical Greeks. Concave dips in the smile produce negative gamma for long options (impossible) or negative local variance (breaks exotic pricing).
Unstable pricing. If moving one market quote causes large changes in interpolated values far away, the surface is noisy. Hedging becomes driven by model artifacts, not market moves.
Methods at a glance
| Method | In a sentence | Good for |
|---|---|---|
| Linear | Straight lines between points | Quick estimates only |
| Cubic Spline | Smooth polynomial curves | Visualization (not production) |
| SVI | 5-parameter model per expiry | Crypto and equity options |
| ORC Wing | Trader-friendly SVI reparameterization | Smile editing by hand |
| SABR | 4-parameter stochastic vol model | Interest rate swaptions |
| Local Vol | Derive instantaneous vol via Dupire | Exotic option pricing |
Hypercall, Deribit, and most crypto vol desks use SVI because it is simple (5 parameters), fast (fits in milliseconds), and can be constrained to prevent arbitrage. For a deep comparison of all methods, see Interpolation Methods.
Step 5: Check for Consistency
Fitting each expiry independently can produce a surface that is internally inconsistent. Before publishing the surface, three constraints must hold.
Calendar arbitrage
Total variance must increase with maturity at every strike. If it does not, you could sell a short-dated straddle and buy a longer-dated one at the same strike, collecting more premium than you pay. Free money.
Calendar Arbitrage Check
Total variance (σ² × T) must increase with maturity at every strike.
Toggle between states. When the 30d total variance dips below 7d, selling a 7d straddle and buying a 30d straddle would be risk-free profit.
Toggle between states: Consistent (each curve sits above the previous), Violation (the 30d curve dips below 7d near ATM), and After Fix (parameters adjusted so the constraint holds).
Butterfly arbitrage
Call prices must be convex in strike. If the smile has a concave dip (a local valley), a butterfly spread centered on that dip produces risk-free profit.
Call spread monotonicity
Call prices must decrease in strike, and the rate of decrease must stay between 0 and 1. A higher strike means a higher hurdle, so the call must always be worth less.
The Result: A Complete Surface
After all five steps (price inversion, shape understanding, gap filling, consistency enforcement), we get a smooth, continuous vol surface. Here it is:
Volatility Surface
Calm markets. Mild put skew, slight contango.
| Strike | 7d | 14d | 30d | 60d | 90d |
|---|---|---|---|---|---|
| $80k | 51% | 52% | 53% | 55% | 57% |
| $85k | 50% | 51% | 52% | 54% | 56% |
| $90k | 50% | 50% | 51% | 54% | 56% |
| $95k | 49% | 49% | 51% | 53% | 55% |
| $100k(ATM) | 48% | 49% | 50% | 52% | 55% |
| $105k | 48% | 49% | 50% | 53% | 55% |
| $110k | 49% | 49% | 50% | 53% | 55% |
| $115k | 49% | 49% | 51% | 53% | 55% |
| $120k | 49% | 50% | 51% | 53% | 55% |
Click expiry headers to isolate a skew slice. Click strikes to see term structure.
In the 3D view, drag to rotate and scroll to zoom. Toggle scenarios to see how different market conditions reshape the surface. In the 2D view, click an expiry header to isolate a skew slice, or click a strike to see its term structure.
The four scenarios show how dramatically the surface changes:
- Normal: Mild put skew, slight contango. The default state.
- Pre-Event: Near-term vol explodes. Both wings elevate because direction is uncertain.
- Crisis: Everything elevated. Extreme put skew. Steep backwardation.
- Euphoria: Call skew emerges. OTM calls get bid. Rare.
How the Surface Moves
The surface is not a static object. It reshapes constantly, and understanding how it moves matters for hedging.
Parallel shift: The entire surface lifts or drops. ATM vol rises 3 points, and roughly every cell moves by a similar amount. This is the largest source of P&L for vega-exposed positions.
Rotation: The front moves one direction while the back moves the other. This is why you cannot simply add the vegas of a 1-month option and a 1-year option. A $100k vega position in 7-day options is far more exposed than $100k vega in 6-month options, because short-dated vol is more reactive.
Shape change: The smile gets steeper, the wings widen, or the skew tilts further. A position flat on level and tilt can still lose money on a shape change.
Localized deformation: A single strike or narrow region moves independently. This happens around large open interest strikes (pin risk), barrier levels, or unusual flow in a specific expiry.
Surface Across Market Regimes
Calm conditions, no major events. Mild put skew, slight contango.
| Strike | 7 DTE | 30 DTE | 90 DTE |
|---|---|---|---|
| 85k (OTM Put) | 58% | 55% | 52% |
| 90k | 54% | 52% | 50% |
| 95k | 51% | 50% | 48% |
| 100k (ATM) | 48% | 48% | 47% |
| 105k | 46% | 47% | 46% |
| 110k | 45% | 46% | 46% |
| 115k (OTM Call) | 44% | 45% | 45% |
What to notice:
- Mild put skew (~14 vol points from put to call)
- Slight contango (far-term slightly lower)
- ATM vol around 48% - typical for calm BTC
Notice the pattern: crisis surfaces are 2-3x the IV of calm markets. Normally puts dominate the skew, but euphoria can flip it. Calm markets produce contango, events produce backwardation. Pre-event surfaces have fat wings on both sides because direction is uncertain.
Build Your Own
Adjust IV values, check for arbitrage, and see the result in real time.
Build Your Own Skew
Calm market, mild put skew
| Strike | Delta | IV(click to edit) |
|---|---|---|
| $80k | 10Δ Put | 67% |
| $85k | 15Δ Put | 62% |
| $90k | 25Δ Put | 57% |
| $95k | 40Δ Put | 53% |
| $100k | ATM | 50% |
| $105k | 40Δ Call | 51% |
| $110k | 25Δ Call | 53% |
| $115k | 15Δ Call | 56% |
| $120k | 10Δ Call | 59% |
Click IV values in the table to edit directly. Invalid configurations will show arbitrage warnings.
Equation Explorer
Plug in values and compute the key quantities used throughout this page: total variance, forward vol, log-moneyness, and BS pricing.
Equation Explorer
Practical Wisdom
A few things to keep in mind when working with vol surfaces:
The surface is a model, not reality. Only a fraction of the grid points come from real trades. When you trade an illiquid strike, check the bid-ask spread. Wide spread = model output. Tight spread = real liquidity.
Wing quotes are uncertain. Deep OTM vol levels depend heavily on the fitting model. Two platforms using different methods will show different wing IVs for the same underlying. ATM Greeks are the most reliable.
Don't compare vegas across maturities. A $100k vega position in 7-day options is drastically more sensitive than $100k vega in 6-month options. Weight your vegas (square root of time is a start, empirical ratios are better) before netting them.
The histogram lies about the skew. You cannot derive the skew from a histogram of past returns. Histograms hide the path: the fact that vol rises in selloffs and compresses in rallies. The skew is about the co-movement of spot and vol along the path, not the shape of the terminal distribution.
Black-Scholes is wrong but useful. Traders chose to warp the vol parameter of a wrong model (creating the surface) rather than adopt a "correct" model with more parameters. Every additional parameter must be estimated, and estimation error compounds. The surface is the market's pragmatic answer.
Model deep-dives: SVI | ORC Wing | SABR | Local Volatility | All Methods Compared
Related: Vol Surface | Implied Volatility | Skew | Term Structure