A Forecast and a Scorecard
Implied volatility and realized volatility answer two different questions about the same thing. Implied volatility (IV) asks: how much movement is the options market pricing in for the future? Realized volatility (RV) asks: how much movement actually happened? One is a forecast embedded in option prices; the other is the scorecard, computed from the underlying's own returns after the fact.
Almost every practical use of these two numbers, deciding whether options look expensive, judging whether a hedge is worth its cost, understanding why a "winning" trade lost money, comes down to comparing the forecast to the scorecard correctly. That comparison is where most explanations stop short. The definitions are easy; the traps are in the measurement windows, the persistent gap between the two numbers, and what that gap does and does not tell you.
If you need the ground-level primer on what IV is and how options encode it, start with our implied volatility guide. This article assumes that base and focuses on the comparison itself.
How Realized Volatility Is Measured
Realized volatility is the one you can compute with a spreadsheet. The standard recipe: take daily logarithmic returns of the underlying over some lookback window, compute their standard deviation, and annualize by multiplying by the square root of 252 (the approximate number of trading days in a year). A stock whose daily returns have a standard deviation of 1% carries an annualized realized volatility of roughly 16%.
That square root of 252 (about 15.9) is why traders use the "rule of 16" as a mental converter: divide an annualized volatility number by 16 to get the daily move it corresponds to. A 32% volatility implies typical daily moves around 2%; an 80% volatility implies about 5%. The rule works in both directions and applies equally to implied and realized figures, which is what makes it useful for comparing them at a glance.
Two measurement choices matter more than most people expect. First, the window: 10-day realized volatility reacts fast but is dominated by whatever happened in the last two weeks, a single large day can swing it violently. A 30-day or 60-day window is smoother but slower to reflect a regime change. Second, the estimator: the standard close-to-close calculation only sees where each day ended. A session that travels 2% peak-to-trough but closes flat contributes almost nothing to close-to-close RV, which is why range-based estimators (using each day's high and low) exist. In an era where index moves increasingly play out intraday and mean-revert by the close, close-to-close realized volatility can understate how violent the tape actually felt.
Where Implied Volatility Comes From
Implied volatility is not computed from the underlying's returns at all. It is backed out of option prices: given an option's market price, the strike, time to expiry, interest rate, and dividend assumptions, a pricing model solves for the volatility number that makes the model price match the market price. The market sets the price; the volatility is what that price implies.
Strictly speaking there is no single implied volatility for a stock, every strike and every expiration carries its own IV, and the differences between them form the volatility skew. When traders quote one IV number for an underlying, they usually mean something like the at-the-money IV for a near-dated expiry, or an index that aggregates across strikes.
The most famous aggregate is the VIX, which Cboe calculates from S&P 500 option prices to represent the market's expectation of 30-day volatility. That makes the VIX the cleanest large-scale example of implied volatility: a single, continuously published number for what the options market expects the S&P 500 to do over the next month. Our VIX guide covers its mechanics and behavior in depth.
The Window Mismatch Most Comparisons Get Wrong
Here is the trap: today's 30-day implied volatility is a forecast of the next 30 days. Trailing 30-day realized volatility measures the last 30 days. Put them side by side (which is exactly what most charting platforms do) and you are comparing a forecast for one period against the outcome of a different, earlier period.
Most of the time the sloppiness is harmless, because volatility is persistent: calm months tend to follow calm months. But at turning points the mismatch is exactly where the information is. The morning after a crash lands in the data, trailing RV is enormous (it contains the crash) while IV may already be falling as the market prices in stabilization. A naive reading says "options are cheap relative to realized." The correct reading is that the two numbers are describing different worlds: one is looking backward at the crash, the other forward past it.
The honest way to score implied volatility is against the realized volatility that followed it: compare the 30-day IV quoted on a given date with the RV computed over the 30 days after that date. Do that consistently and a striking pattern emerges.
The Volatility Risk Premium
Scored properly, implied volatility does not just track future realized volatility, it systematically overshoots it. A CFA Institute analysis comparing the VIX against subsequent 30-day realized volatility of the S&P 500 across roughly 35 years of data beginning in 1990 found the VIX averaged about 19.6 while forward realized volatility averaged about 15.5, a gap of roughly 4 percentage points, sustained across the full sample. In quiet stretches the overshoot ran even wider: through 1990–1996 the VIX exceeded subsequent realized volatility by roughly five to seven points.
That persistent gap is the volatility risk premium, and it is not a market error. Option sellers are underwriting insurance against large moves: they collect steady premium and occasionally absorb severe, fast losses when volatility explodes. As AQR's research on the volatility risk premium frames it, the premium is compensation for bearing exactly that risk, sellers demand a margin above expected realized volatility for the same reason an insurer charges more than the actuarially expected loss.
The exceptions prove the point. At the onset of genuine crises (2008, the COVID crash of 2020) realized volatility spiked above what the VIX had been pricing, and the insurance sellers paid out. Years of a few points of collected premium can be handed back in weeks. The premium exists precisely because those episodes are real.
Reading the Spread Like a Trader
The IV-versus-RV spread compresses a lot of market information into one number, and it is worth reading in both directions.
- IV far above recent RV. The options market is paying up for movement the tape has not yet delivered. Common around scheduled events, ahead of earnings, single-name IV inflates to price the expected gap, which is the "implied move" logic covered in our earnings options guide, and after shocks, when the memory of violence keeps insurance bid even as the tape calms. Options are expensive in realized terms; buyers need the future to be wilder than the recent past just to break even.
- IV near or below recent RV. Rare and worth attention. Either the market is confident the movement just witnessed will not continue (post-event IV crush is the classic case), or realized volatility is erupting faster than the options market can reprice, the crisis-onset signature from 2008 and 2020.
- The ratio over time. Because the premium is persistent, IV divided by trailing RV spends most of its life above 1. The informative moments are the extremes: an unusually stretched ratio says option prices embed a lot of fear per unit of delivered movement; a ratio pinned near or below 1 says the market is struggling to keep up with the tape.
One caveat: a 30-day IV number and a 30-day RV number still say nothing about how movement is distributed across the curve. Front-month and back-month implied volatilities routinely disagree, and that shape, covered in our VIX term structure guide, often carries more regime information than any single spread reading. The live VIX futures curve is where to check it against today's prints.
Limits of the Comparison
Every IV-versus-RV comparison inherits the measurement choices underneath it. Which implied volatility, at-the-money, or a strike out on the skew? Which realized estimator (close-to-close, or range-based? Which window) 10 days of noise, or 60 days of lag? Change the inputs and the "spread" can shrink, grow, or flip sign without the market changing at all.
None of this makes the comparison useless. It makes it a diagnostic rather than a signal: a way of asking whether the price of insurance is high or low relative to the weather actually delivered. The forecast is usually pessimistic on purpose (that is the premium) and the scorecard occasionally, briefly, catastrophically exceeds it. Understanding both halves, and the windows they are measured over, is what separates using these numbers from being used by them.
Educational content, not financial advice. See our risk disclosure.