Historical vs Implied Volatility

Backward-looking versus forward-looking volatility, compared — what each measures, why they diverge, and how traders use the spread.

The distinction between historical vs implied volatility sits at the center of almost every options strategy. Historical volatility tells you how much an asset has moved; implied volatility tells you how much the options market expects it to move. They are measuring different things, derived from different inputs, and they diverge in systematic, exploitable ways. Getting them confused — or treating them as interchangeable — is one of the more expensive beginner mistakes in options trading.

Historical volatility: the formula and what it measures

Historical volatility (HV), also called realized volatility, is the annualized standard deviation of a security's log returns over a lookback window. The formula is:

σ = stdev(ln(P_t/P_t−1)) × √252

where the stdev is computed over n daily closing prices, ln(P_t/P_t−1) is the daily log return, and √252 annualizes from daily to annual (252 trading days per year). A 20-day HV uses the most recent 20 daily returns. A 252-day HV uses a full year.

The lookback window matters enormously. A 10-day HV is a noise-heavy snapshot of recent choppiness. A 252-day HV is a smoothed, slow-moving average that lags every regime change. Most practitioners watch several windows simultaneously — 10-day, 20-day, and 60-day — and weight them by what the trade horizon demands. The common shorthand is that 20-day HV is the standard comparator for near-term options.

A worked calculation

Suppose an ETF closes at 100, 101, 99, 102, 103, 101 over five days. The four daily log returns are:

• ln(101/100) = 0.00995
• ln(99/101) = −0.01980
• ln(102/99) = 0.02985
• ln(103/102) = 0.00980
• ln(101/103) = −0.01942

The sample standard deviation of those five returns is roughly 0.01853. Annualized: 0.01853 × √252 ≈ 0.2940, or about 29.4% HV. That is a moderately volatile name — nothing extreme, but well above the 13–16% range that large-cap indices print in low-vol regimes.

Implied volatility: the market's forward number

Implied volatility is derived differently. Rather than computing anything from the underlying's price history, you take a live option price, plug it into a pricing model (usually Black-Scholes or a close variant), and solve backward for the single volatility input that makes the model price match the market price. That recovered σ is the IV.

IV is forward-looking by construction: the option market is pricing the distribution of returns from now to expiration, not over the past 20 days. An earnings event six days out will inflate near-term IV even if the stock has been dead quiet for three months — because the event risk is ahead, not behind. Use the options pricing calculator to see how plugging in different volatility inputs moves the theoretical price, and get a feel for how sensitive IV extraction is to the model you choose.

IV is also strike-specific. A 25-delta put and a 25-delta call on the same expiration will typically imply different volatilities — the volatility smile — which tells you the market assigns non-lognormal probabilities to tail moves. When people refer to a stock's IV, they usually mean the at-the-money IV for the nearest liquid expiration.

The variance risk premium: why IV usually exceeds HV

Across most assets and most time periods, implied volatility runs above subsequent realized volatility. This gap is the variance risk premium (VRP). The intuition is straightforward: option sellers bear the risk that realized vol explodes, so they demand a premium to hold short vol. Option buyers, who are effectively buying insurance against big moves, pay that premium. The result is that IV embeds a markup over the market's unbiased expectation of future realized vol.

Empirically, the VRP on equity index options has historically averaged somewhere between 2 and 5 vol points on the S&P 500 — meaning if the market expected 15% realized vol, it would price options at roughly 17–20% IV. The premium widens dramatically in stress periods and compresses (sometimes inverts) when realized vol spikes unexpectedly above what the market had priced.

The VRP is also why short-vol strategies have historically generated positive expected returns on a raw basis. Selling options is not free alpha — you are being paid to absorb variance risk, and that risk materializes in violent drawdowns during volatility regimes like March 2020 or August 2015.

IV rank and IV percentile

Raw IV levels are hard to interpret in isolation. A 30% IV on a biotech stock may be cheap; a 30% IV on a utility may be the highest it has ever been. IV rank and IV percentile solve this by contextualizing current IV against its own history.

• IV rank (IVR) is defined as (current IV − 52-week low IV) / (52-week high IV − 52-week low IV), expressed as a percentage. An IVR of 80 means current IV is 80% of the way from its annual low to its annual high.
• IV percentile is the percentage of trading days over the past year on which IV closed below the current level. An IV percentile of 80 means IV was lower than today on 80% of days in the past year.

The two numbers often differ. IVR is sensitive to a single extreme spike (a high-IV day compresses every subsequent IVR reading). IV percentile is more robust to outliers. Most volatility desks track both. An IVR above 50–60 is typically the threshold where option-selling strategies start to look attractive on a risk/reward basis.

Using the HV/IV spread as a signal

The spread between IV and recent HV is the most direct measure of whether options are cheap or rich. If 20-day HV is 18% and 30-day IV is 26%, options are pricing in significantly more volatility than the stock has recently delivered — they are rich by 8 vol points. If HV is 24% and IV is 19%, the market is underpricing recent turbulence relative to what the stock is actually doing — options look cheap.

Condition	Signal	Typical trade bias
IV > HV by >5 pts	Options rich	Net short vega: sell premium, spreads, iron condors
IV ≈ HV	Fair value	Direction-driven; vol less of a tailwind either way
IV < HV by >3 pts	Options cheap	Long vega: buy options, straddles, or calendar spreads

Signals derived from the HV/IV spread are only as reliable as the underlying data. Stale option quotes, bid/ask midpoint errors, and corporate-action distortions in historical price series all corrupt the calculation. This is why access to clean, adjusted, tick-quality data from reputable volatility data providers is a practical prerequisite for running a systematic vol strategy rather than a nice-to-have.