Implied Volatility: A Price, Not a Forecast

Michael Whitman · Trading since 2017, full-time since 2020 · Published 2026-09-06 · Updated 2026-10-19 · Reviewed 2026-10-03

Every figure on this page comes from real option quotes: Cboe’s delayed quotes for ten US tickers, saved after the close on 25 September 2026. Nothing here is a forecast of anything.

How it works

An option pricing model takes several inputs — price, strike, time, rates and expected volatility — and produces a value. Four of those are known. Volatility is not.

So the calculation is run backwards. Take the price the option is actually trading at, and solve for the volatility figure that would produce it. That figure is the implied volatility.

Which means it is a price with the units of a forecast. It is set by what people are paying, not by anything about the future, and calling it “the market’s expectation” is a description of who set it rather than evidence that they were right.

The event cycle

Before a scheduled event, implied volatility rises. Earnings, a rate decision, a trial result — anything with a known date and an unknown outcome makes a large move more likely, and options get more expensive.

Once the result is known, it usually drops. The uncertainty that justified the price has been resolved, and the premium falls whichever way the news went. That drop is what traders mean by “vol crush,” and the IV crush page measures it around real earnings dates.

Which produces the standard beginner experience. Buy calls before earnings, be right about the direction, and lose money — because the fall in implied volatility took more than the move gave. The share was up and the option was down, and nothing about that is unusual.

It is separate from theta and works alongside it. Time decay removes value on a schedule; a fall in implied volatility removes value in one step. A position can suffer both at once, which is what the week after earnings looks like.

A worked example

Take SPY’s options expiring on 20 November 2026, 56 calendar days after the 25 September close of $771.35. The listed strike nearest the close was $771. Cboe priced its call at 13.53% and its put at 13.49%, a mean of 13.51%.

Turn that into a move. Implied volatility is quoted per year, so scale it by the square root of the time: 13.51% × √(56 / 365) = 5.29%. On $771.35 that is $40.82. Under the model’s own assumptions, that is one standard deviation of movement by expiry, the size of swing the option prices were paying for.

Now go 10% down the chain. The strike nearest 90% of the close was $695, 9.90% below it, and its put was priced at 21.39%, almost eight points above the at-the-money figure. The call at $850, 10.20% above the close, carried 11.92%. At the mid-point between bid and ask the $695 put cost $2.67 and the $850 call $0.375, about a seventh of the price, for a strike almost the same distance away.

That gap is the market charging more for crash protection than for a rally. The model treats both strikes with one volatility; the prices refuse to, and solving each price separately is what exposes it.

The original data

The snapshot. Cboe’s delayed quotes as saved after the close on 25 September 2026: 13,290 SPY contracts plus full chains for QQQ, IWM, AAPL, MSFT, NVDA, AMZN, META, TSLA and AMD, each carrying Cboe’s own implied volatility per contract and its 30-day figure for the underlying.

One expiry, 38 strikes. For the SPY 20 November 2026 expiry, every $5 strike from 85% to 110% of the close was taken from the side that is out of the money: puts below $771.35, calls at or above it. The figure ran from 25.55% at the $660 put down to 11.47% at the $825 call, then edged back up to 11.92% at $850.

The same tilt showed up on all ten tickers, at very different strength. Using the same expiry, the put nearest 10% below each close was priced above the at-the-money figure every time. The gap was widest on the three funds — SPY +7.88 points, QQQ +6.07, IWM +5.57 — and narrowest on single stocks: AAPL +3.29, NVDA +2.09, MSFT +1.59, AMZN +1.15, TSLA +0.78, META +0.23 and AMD +0.17.

The level itself changed more than four times over. Cboe’s 30-day figure was 11.98% for SPY, 17.81% for IWM, 18.05% for QQQ, 22.22% for AAPL, 28.22% for MSFT, 30.12% for NVDA, 31.92% for AMZN, 41.22% for META, 43.58% for TSLA and 51.49% for AMD. A reading of 30% was low for one of these and high for another on the same evening.

Then set each against what actually happened. Realized movement here is the annualized standard deviation of the 21 daily close-to-close log changes between the 26 August and 25 September 2026 closes. The 30-day implied figure sat above it for seven tickers, from TSLA at 43.58% against 43.52% to IWM at 17.81% against 12.68%. It sat below for three: NVDA at 30.12% against 43.11%, META at 41.22% against 49.28% and AMD at 51.49% against 54.00%. Realized movement for the other five was 10.73% for SPY, 15.82% for QQQ, 21.65% for AAPL, 25.08% for MSFT and 27.26% for AMZN.

What this does and does not show. It is one evening’s prices, not a history, so it cannot say how often implied ends up above what later arrives. It does show the three claims this page makes: one contract chain holds many implied volatilities, the raw level means little across tickers, and the price of options can sit well away from recent movement in either direction. The rows behind the chart are in the ten-ticker file and the SPY strike file.

In practice: reading the number

A raw figure means very little. As the ten tickers show, 30% is low on one instrument and high on another. What matters is where it sits against that instrument’s own history — which is what percentile-style measures exist to express.

It feeds every other number. Delta, gamma and theta are all computed with it as an input, so a change in implied volatility changes the whole profile of a position without price moving at all.

Cheap options in a quiet market are usually cheap for a reason. Low implied volatility reflects a market that has not been moving, and buying options because they are inexpensive is buying the market’s assessment that nothing is likely to happen.

What implied volatility is not

It is not a forecast. It is what people paid, and a price is not a claim that can come true or false.

It is not directional. It rises when large moves are expected in either direction, so a high figure says nothing about which way.

It is not the same as realized volatility. How much the instrument actually moved is a separate, measurable quantity, and on 25 September 2026 the two differed by as much as 12.99 points (NVDA).

And it is not comparable across instruments. A number that is extreme for one asset is ordinary for another, which is why the raw figure is close to useless without its own context.

How this connects

Plotted against strike, it becomes the volatility smile. On index options the shape is usually a smirk, tilted so that puts below the market carry the higher figure — the SPY curve above is a measured example.

The gap to realized movement has a name. When implied sits above what later arrives, the difference is the volatility risk premium, or the variance risk premium when measured in squared terms.

Strategies are really positions on this number. Buying a straddle or strangle is long it; a covered call, a cash-secured put, a credit spread or an iron condor is short it, whatever the directional story.

The VIX is the same idea for the S&P 500 index as a whole, which Cboe describes as read from S&P 500 index option prices rather than from one contract.

And chart tools measure the realized side. Average true range and the Bollinger Band width track how much price has actually moved, and the squeeze describes the quiet periods that often precede a jump.

When it fails

The core error is treating it as information about the future. It is information about what options cost today, and the market setting it is no better informed than any other market.

The second is buying options into a known event. The premium already contains the expected move, so the position needs a move larger than the one everybody is expecting — which is a much harder bar than being right about direction.

A third is selling volatility because it looks high without asking why. It is frequently high because something genuinely uncertain is about to happen, and the seller is being paid for a real risk rather than a mispricing. On 25 September 2026 three of ten tickers had moved more over the past month than their options were pricing, so “high” and “rich” are not the same thing.

A fourth is reading one strike as the whole chain. Quoting “SPY implied volatility” without a strike hides a range from 11.47% to 25.55% inside one expiry, and a put buyer pays the top end of it.

A fifth is ignoring the transaction costs around it. Option spreads tend to widen when volatility rises, so trading it can be most expensive exactly when it is most interesting — on top of the spread the underlying already charges.

And a sixth is assuming the crush is a free trade. Selling into an event to capture the drop means being short the move itself, which is the one thing the elevated premium was pricing.

Nothing here says the number is useless. It converts an option price into something comparable across strikes and dates, which a dollar premium cannot do. What it does not do is tell you what will happen — and separating those two claims is most of what it takes to use options without being surprised by them.

Related

Vega is what a change in this number is worth to a position. The volatility smile is the pattern the SPY strikes above trace out. Options expiry is the other clock running alongside it, and put options are where the higher downside figure is paid.

This page is educational, not financial advice. Test every idea on your own charts before risking money.