Implied Volatility

What does implied volatility actually measure?

Spot and option premiums are quoted, while strike and expiration are contract terms. The pricing model also uses a selected rate curve, expected dividends and an exercise convention. Future realized volatility is unknown; an implied-volatility calculation solves for the volatility that matches the premium under those assumptions.

So the market solves it in reverse. Supply and demand set the premium, and implied volatility is the volatility number that makes the model reproduce that premium exactly. It is solved from the observed price and then can be used as an input when repricing scenarios. When a platform displays "IV 34%" next to a contract, it is reporting the answer to the question: what annualized volatility would justify what this option just traded for?

Two properties follow immediately, and both get misread constantly.

First, implied volatility is annualized regardless of how long the option has left. A weekly contract and a two-year contract both quote in annualized percentage points, which is what makes them comparable. It is also why a 65% weekly reading looks alarming: when an earnings report falls inside the week, that one overnight jump is annualized into the headline number.

Second, it is a magnitude, not a direction. Implied volatility widens the distribution the model considers possible in both directions at once. This is the same property that makes market volatility directionless: a smooth 10% rally and a violent round trip back to unchanged can produce very different volatility readings, and the higher one belongs to the trade that went nowhere.

Vega translates an implied-volatility change into option value. A vega of 0.20 per volatility point gives a $20 value change per standard contract for a one-point IV move. Long vanilla calls and puts carry positive vega; short positions carry the opposite exposure. Multiply each contract’s own vega by the IV change and position size, then combine the legs.

How does implied volatility compare with realized volatility?

Realized volatility is the measured counterpart. It takes a completed series of returns, computes their standard deviation, and annualizes it by the square root of the number of periods in a year, commonly 252 trading days for US equities. Five daily log returns of +1.0%, −0.5%, +1.5%, −1.0%, and +0.5% have a sample standard deviation of about 1.037% per day, which annualizes to 1.037% × the square root of 252, or 16.46%. The annualization window and variance estimator are choices; Cboe’s realized-volatility methodology specifies the conventions for its own indices.

The comparison between the two is the central trade in volatility. Implied volatility is what the market charged in advance. Realized volatility is what the underlying delivered. Implied volatility minus the realized volatility that follows over the same horizon measures the volatility risk premium. An unhedged option’s cash result also turns on where the stock finishes, so judge the premium by that spread and the trade by its payoff.

The long-run average has favored the seller. Cboe’s May 2019 summary of Bondarenko’s study reported average VIX of 19.3% against 15.1% average realized volatility over 1990–2018, a 4.2-percentage-point volatility spread. The premium exists because option sellers take open-ended exposure to sudden large moves and price compensation for it, while portfolio holders pay standing demand for downside protection.

The sign is an average, not a rule. The premium inverts hard during shocks, when realized volatility outruns the implied level that was sold, and those episodes are exactly where the long-run compensation gets paid out. The implied volatility vs realized volatility comparison covers the horizon-matching problem in detail: set a 30-day implied reading against 30 days of realized volatility, because a five-day realized spike measures a different horizon.

How do IV rank and IV percentile normalize a raw quote?

An absolute implied volatility number gets its meaning from the underlying’s own history. Forty percent is placid for a small-cap biotechnology name and near-record for a utility. Two normalizations place the number in that history.

Implied volatility rank measures distance through a range:

IV rank = 100 × (current IV - lookback minimum) / (lookback maximum - lookback minimum)

Implied volatility percentile measures frequency:

IV percentile = 100 × (observations below current IV) / total observations

Take a hypothetical MNO with 30-day constant-maturity at-the-money IV of 44%, a 52-week low of 26%, and a 52-week high of 62%. Distance above the low is 44 minus 26, or 18 points. The full range is 62 minus 26, or 36 points. IV rank is 100 × 18 / 36, which is 50. Now count the daily observations: 189 of the 252 trading days in the window closed below 44%. IV percentile is 100 × 189 / 252, which is 75.

Rank 50 places the current 44% IV halfway between the 26% low and 62% high. Percentile 75 places it above 189 of the 252 sampled readings. Use both to compare location within the range with frequency in the history, then read the full series to see how the observations are distributed.

Match the inputs before you compare a rank or percentile from another platform. MNO’s 50 and 75 come from 30-day constant-maturity at-the-money IV over a 52-week window; build them from a chain-wide composite or a different lookback and the same formulas return different readings.

How do you turn implied volatility into an expected move?

Implied volatility is annualized, so converting it to a shorter window means scaling by the square root of time. The expected move formula is:

Expected move = underlying price x implied volatility (as a decimal) × square root of time in years

A $120 stock at 36% implied volatility with 30 calendar days remaining gives time in years of 30 / 365 = 0.0822, and an expected move of 120 × 0.36 × the square root of 0.0822, which is $12.38. That frames a one-standard-deviation band of $107.62 to $132.38 under a calendar-day convention.

The shortcut version is the rule of 16. The square root of 252 is about 15.87, so annualized IV divided by 16 approximates a one-day move. A VIX print of 20 implies roughly 20 / 16 = 1.25% for a one-standard-deviation session, which on an S&P 500 near 5,000 is about 62.5 index points. The same annualized 20 divided by the square root of 12 gives about 5.8% for a month.

Match the day count before comparing two expected moves: the $12.38 uses 365 calendar days, and a 252 trading-day denominator returns a different figure from the same inputs. The at-the-money straddle price is the other expected-move convention. It is the premium a long straddle must recover, so it sets that trade’s break-even distance at expiration. And plan for closes outside the band: the familiar 68% coverage assumes a normal distribution, while real returns jump and change volatility.

Why is the volatility surface not flat?

If one volatility priced every contract on an underlying, a chain would show the same IV at every strike and every expiration. It never does.

Read implied volatility horizontally, across strikes within one expiration, and the result is volatility skew. Equity and index chains commonly show downside skew: on one 45-day expiration, a 25-delta put might print 31% IV, the at-the-money option 26%, and the 25-delta call 24%. That is 7 volatility points of put-versus-call skew and 5 points of put-versus-at-the-money premium. Standing demand for portfolio protection, jump risk, and the cost of hedging an asymmetric distribution all show up in that shape.

When both wings lift above the center rather than one, the curve is a volatility smile. A 45-day expiration printing 28% at the 90 strike, 20% at the 100 at-the-money strike, and 27% at the 110 strike has curvature of 27.5% minus 20%, or 7.5 volatility points, with only 1 point of wing asymmetry. That is a genuinely two-sided curve, common ahead of binary outcomes where both tails are live. The volatility skew vs volatility smile comparison separates the vocabulary, which traders routinely blur.

Read implied volatility vertically instead, across expirations at matched delta or moneyness, and the result is volatility term structure. A chain quoting 52% at 14 days, 38% at 45 days, and 34% at 90 days has a 45-day minus 14-day spread of −14 points and a 90-day minus 45-day spread of −4 points. That front-loaded inversion points to a catalyst inside the nearest expiration, so check the event calendar before trading the front month.

Stack every strike slice across every expiration and the result is the volatility surface. Take a fitted surface showing 32% / 25% / 23% at 90%, 100%, and 110% moneyness for 30 days, against 29% / 24% / 22% for the same moneyness points at 90 days. Downside skew exists in both tenors, steeper in the front, while the at-the-money term spread is only 1 point. The headline IV is one point on that surface; price a spread from the strikes and expirations it actually holds.

One arithmetic rule governs all of it: variance, not volatility, is additive through time. Total implied variance is IV squared times years to expiration. Subtracting two annualized IV percentages across different maturities is not a valid decomposition, which matters enormously for the next section.

What does the earnings cycle do to implied volatility?

The earnings volatility cycle is the clearest case of concentrated event variance moving into option prices and then out of them. Here is a fully worked version.

Setup. MNO trades at $100.00 and reports earnings in six days. Expiration A has 5 days remaining and ends before the report, quoting 32% at-the-money implied volatility. Expiration B has 12 days remaining, contains the report, and quotes 58%. The calls are European, priced with Black-Scholes on calendar days / 365, zero rates and dividends, and 100-share contracts. A’s 32% sets the rate for B’s ordinary days, so whatever B charges above it is the event.

Step 1, convert both quotes to total implied variance.

• Expiration A: 0.32 squared x 5 / 365 = 0.1024 × 0.01370 = 0.001403
• Expiration B: 0.58 squared x 12 / 365 = 0.3364 × 0.03288 = 0.01106

Step 2, strip out the ordinary-day baseline. Carrying A’s 32% baseline across B’s full 12 days gives 0.1024 × 0.03288 = 0.003367 of baseline variance. The event component is 0.01106 minus 0.003367 = 0.007693.

Step 3, express the event as a price distance. The square root of 0.007693 is 0.0877, so the market is pricing an event-only move of 8.77%, or $8.77 on a $100 stock. The full 12-day expected move is 100 × 0.58 × the square root of 0.03288 = $10.52. Those reconcile because variances add: 5.80 squared plus 8.77 squared equals 110.6, whose square root is 10.52. The rule-of-16 cross-check agrees. Count the 12 days as eight trading sessions: 58 / 16 is 3.6% per session, and 3.6% × the square root of 8 gives 10.2%.

Step 4, price the at-the-money call in expiration B. At $100 spot, a 100-strike call at 58% IV with 12 days left is worth $4.19 per share, or $419 per contract. Its vega is 0.072, or $7.20 per contract per volatility point.

Step 5, apply the crush. Six days pass, the report lands, the event variance leaves the contract, and expiration B resets to the 32% baseline with 6 days remaining. That is a 26-point drop in implied volatility. The linear vega estimate is 0.072 × 26 = $1.87 per share, or $187 per contract. Full repricing gives more: with the stock unchanged, the call is worth $1.64, a decline of $2.55 per share or $255 per contract. The two pieces separate cleanly. Holding implied volatility at 58% while the six days pass takes the call from $4.19 to $2.97, so decay accounts for $1.22. Repricing that six-day contract from 58% to 32% takes it from $2.97 to $1.64, so the crush itself accounts for $1.33. The crush term lands below the $1.87 linear figure because vega scales with the square root of remaining time, so the 12-day vega overstates what the 6-day contract actually loses per point. That is why a large IV move calls for a full repricing rather than the linear vega figure.

Step 6, run the direction scenarios. Repricing the same 100-strike call at 32% IV with 6 days remaining across post-report outcomes:

The call needs a 3.8% rally just to return the $419 paid. A correct directional call of +3.0% still loses money. That is implied volatility crush in one line: the event premium was charged up front, so the stock has to clear a hurdle set by what was already priced before the buyer gains anything. Note also that the break-even hurdle of 3.8% sits well below the 8.77% event-implied move, because delta gains accrue on the way there. Treating the implied move as a breakeven is a separate error from ignoring the crush entirely.

What does implied volatility look like on a live tape?

On a live chain, implied volatility is a column, not a number. Each strike carries its own solved value, so scanning down the puts and up the calls at one expiration draws the skew by hand. Scanning the same delta across the expiration dropdown draws the term structure. The at-the-money row of the first expiration after a scheduled report is where event premium concentrates, and its distance above the expiration that ends before the report is the raw material for the variance decomposition above.

Wide bid-ask spreads change what the number means. Implied volatility is solved from a price, so an illiquid wing with a $0.05 by $0.35 market has an interval of plausible IV rather than a point. A far out-of-the-money contract can show a dramatic IV that is untradeable near the midpoint. That is the job of fitted surfaces and quote filters. Before trading a screaming skew reading on a thin chain, price it from the bid and ask you can actually hit.

Options flow gives the surface a cause. Aggressive wing buying steepens a local part of the curve before at-the-money IV moves at all, so a trader watching only the headline number can miss that the strike they intended to buy became expensive relative to the strike they intended to sell. Paired call and put prints at the same strike often represent a volatility trade rather than two conflicting directional bets, and a large single-leg call print may be one leg of a spread, a collar, or a stock replacement.

At the index level, the same reading is packaged. The VIX aggregates a wide strip of out-of-the-money SPX puts and calls across two expirations bracketing 30 days into a single model-free estimate of expected 30-day variance, annualized into percentage points. Because the index is a calculation and not a security, exposure runs through VIX futures, which settle in cash against a Special Opening Quotation of the index and carry a 1,000 multiplier, so one index point is $1,000 per contract.

The listed VIX futures form a curve across settlement dates. Contango places deferred contracts above nearer contracts, while backwardation reverses that order. A 16.50 front contract and an 18.00 second contract have a 1.50-point spread. Closing the front and opening the second changes the contract held; subsequent repricing drives profit and loss. If the new contract falls from 18.00 to 16.50, the loss is $1,500 per standard VX contract. Track each entry, exit, multiplier, and contract weight to calculate the roll strategy’s return.

What are the standard mistakes, and why is each one wrong?

Treating a high IV number as expensive. Read IV against the underlying’s own history and its current catalyst calendar, because forty percent is placid for one name and extreme for another. IV rank and IV percentile supply that history, and reading both shows the shape of the distribution behind the number.

Treating an IV spread as a difference in total variance. Convert each reading to IV squared times time before comparing the variance priced into two horizons. A 52% two-week reading and a 34% ninety-day reading have an 18-point annualized IV spread, and that spread describes the shape of the curve. Variance is what accumulates with time, so the two-week contract carries far less total variance despite the far higher headline.

Reading the expected move as a boundary. Size a short-premium position to survive a close beyond the band. The expected move is a price, and it carries the volatility risk premium, crash-protection demand, dealer inventory, rates, dividends and the spread. The realized move lands inside or outside it, and real return tails are fatter than the normal curve behind the one-standard-deviation label.

Buying an event option on a correct directional view. Price the hurdle before you buy: the move has to clear the event premium already charged. In the worked example above, the $419 call needed a 3.8% rally just to hold its value once the 26-point crush and six days of decay landed.

Assuming the volatility risk premium is free money. Size short-volatility positions for the shock, not the average month. The premium is earned in small frequent increments and surrendered in rare violent episodes, and a strategy that looks smooth for months can return a year of gains in one shock. That shock is precisely the risk the premium compensates.

Confusing spot VIX with the instruments that track it. Price the trade off the future you hold, not the spot index. Futures price expected volatility as of their own settlement date, so a front-month contract can gain only a fraction of a spot move that a trader called correctly. Exchange-traded products hold and roll those futures and inherit the roll cost of the curve.

Applying one parallel IV shock to a multi-leg position. Run skew steepening, wing flattening, and front-end event compression as separate scenarios. A calendar, diagonal, butterfly, or ratio spread occupies several coordinates on the volatility surface, and those coordinates move independently, so a single vega number cannot represent them.

What to learn next

The natural next step depends on which dimension of the surface is doing the work in a given position.

For the Greek mechanics of volatility exposure, vega is the direct sensitivity, and the theta vs vega comparison separates decay from volatility repricing, which are the two forces acting on extrinsic value at the same time and in the same direction after an event.

For the strike dimension, volatility skew and volatility smile explain why two spreads of identical width can have very different economics, and why risk reversals and collars are effectively skew trades with a directional side effect.

For the time dimension, volatility term structure and the earnings volatility cycle cover forward variance and event isolation, the tools that make calendar spreads and event trades measurable rather than intuitive.

For the regime backdrop, market volatility, the VIX, and VIX futures frame whether the tape is in the contango default or an inverted stress state, which changes what a given premium level is actually compensating.