IV Rank: When Premium Is Rich

Relative height, and the expected move

Implied volatility is the market’s price for movement over the life of the contract, so a bare number is uninterpretable: 30% is high for a utility and low for a biotech. IV rank fixes that by placing today inside the instrument’s own past year — the current value less the year’s low, divided by the range — which turns a price into a position in a range. High rank means premium is expensive relative to this stock’s own history, which is when selling is being paid and buying is expensive; low rank is the reverse. Percentile is a close relative that counts the share of days below today’s value instead of the distance between the extremes, and the two usually agree closely enough that either serves. The second tool is the expected move: implied volatility times the square root of the time fraction, which converts the price back into a distance. A 48% IV with thirty days to run implies a move of about 13.8%, so the market is charging for a band of roughly that size around the current price. That is a checkable statement, and it is what turns a vague sense that options are expensive into a comparison: if your own work says the coming event moves the stock 5% and the priced move is 13.8%, the premium is rich relative to your view and the seller has the better side. From a price to a distance — IV low 18%, high 62%, now 48%: Rank 68% — upper two-thirds of the past year · IV at the 50% rank: 40% — the level below which premium selling stops being paid · 48% IV, 30 days: Expected move ≈ 13.8% ← · A trader’s own view: a 5% event: Premium rich — the seller has the better side of that disagreement ← The expected move is one standard deviation, so a third of outcomes land outside it by construction. A band is not a prediction of the range; it is a statement about the distribution that the premium is charging for.

Where implied volatility comes from, and why it is usually above realised

Implied volatility is solved from the premium, not observed, which means it is the market’s price for movement rather than a measurement of past movement. Realised volatility is the same quantity computed from what actually happened, and the two differ in a systematic way: implied is usually above subsequently realised, because sellers demand compensation for carrying the risk that movement arrives suddenly and in one direction. That gap is the volatility risk premium, and it is the reason premium selling has a positive average return and occasional very large losses — a small, steady edge punctuated by the tail. Two practical consequences follow. First, the comparison to make is not whether IV is high in absolute terms but whether it is high relative to realised and to the instrument’s own range, because that is where the compensation is. Second, the same premium that pays steadily is the one that hurts in a shock: an event that produces a 4-sigma move costs the seller far more than the accumulated credit, which is why the position has to be sized for the tail rather than for the average week. • IV rank = (current − low) ÷ (high − low): a price placed in its own range. • High rank favours selling premium; low rank favours buying it. • Expected move = IV × √(time fraction) — one standard deviation, not a forecast. • Implied usually exceeds realised, which is the volatility risk premium. • That premium is paid steadily and lost in the tail, so size for the shock. IV rank is bounded by the past year, so a stock whose business changed will answer a question about its own history rather than about its future. A name that has permanently rerated from a 15% IV regime to a 40% one can show a high rank for months, and the rank will keep saying expensive while the premium is merely appropriate.

Realised volatility is measured, not felt

Almost every learner describes volatility by how it feels — “this thing moves 3% a day”. That feeling is measurable, and the conversion is worth carrying because it is the only way to compare the premium to what actually happened. Take the daily percentage moves, square them, average them, take the square root, and multiply by the square root of 252 trading days. A stalwart that moves 1% a day on average annualises to about **16%**; a name that swings 2% a day is around **32%**; a 4% mover is above **60%**. With that in hand the comparison the lesson is built on becomes arithmetic rather than impression: if the chain is charging 48% and the stock has been delivering 32%, the seller is being paid 16 points a year to take the other side. That is the volatility risk premium in its raw form, and it is also why the same trade loses badly in the weeks when realised runs above implied — the two numbers are not bound to each other, they are only related over long samples. One caveat that separates a measured number from an honest one: close-to-close volatility ignores the move overnight and the path within the day, so it understates what a stop-loss rule will actually experience. If you are sizing positions around stops, measure the intraday range instead.

The blind spot inside a 52-week rank

IV rank compares today’s implied volatility with the highest and lowest implied volatility of the past year, so it is a **percentile of a range** — and a range is only meaningful while the sample still describes the instrument. When a name changes character inside that window, the rank stops describing anything real. A biotech that read 25% implied for nine months and then went to 90% after a pivotal trial will show a rank near zero six months later, because the top of the range is now a number the market has no intention of charging again. The rank reads “cheap” while the option is priced for an event that has not happened yet. The repair is to stop asking how this volatility compares with its own past and start asking how it compares with the movement that is actually arriving. Two substitutes do most of the work. **IV percentile** is the share of days in the window that closed below today’s level, rather than the position within the high-low range, and it is far less sensitive to a single outlier spike. And the **implied-versus-realised spread** — current implied minus the realised volatility of the last twenty or thirty days — answers the question a premium seller actually cares about, which is whether they are being paid more than the movement they are taking on. One further caveat belongs here because it produces a lot of bad trades. A rank is computed from a single tenor, usually around thirty days, so a name whose term structure is steep can be genuinely expensive at one tenor and cheap at another. Before treating a high rank as a signal, look at the whole curve: a company with earnings already inside the thirty-day window will show elevated short-dated implied volatility and a lower rank further out, and selling the short-dated premium because the rank is high means selling straight into the event that created it. Rank is a level, not a forecast, and it is not even a good level when the instrument’s volatility regime has moved. Use it to filter a list of candidates, then decide on the implied-versus-realised spread and the term structure.

The term structure, and how much of the premium is the event

A single implied volatility number is an average across strikes and dates, and the dates part of it is separately informative. Line up the implied volatility of every expiry on the same name and you have the term structure, and its shape says what kind of uncertainty the market is charging for. A flat, gently rising curve is a normal market: movement is expected to be roughly what it has been, with a small premium for the further horizon. A steeply falling curve — high at the front, much lower later — is an event curve, and it means a scheduled uncertainty sits inside the near expiry. A curve high at both ends and lower in the middle is rarer, and usually reflects a dated event plus a broader concern about the medium term. The event curve is where the arithmetic pays. If the thirty-day expiry is charging 65% and the expiry beyond the event is charging 30%, almost all of the difference is the event, and because the event falls on a known date the extra variance can be isolated: the event contributes roughly the difference in variance scaled by the fraction of the period it occupies. On a name reading 65 at the front and 30 behind, the gap implies the market is pricing a move of about twelve percent across the event alone — a number you can argue with, and the same quantity a straddle price approximates, obtained from two implied volatilities instead of one option price. Two readings follow. The first is diagnostic: before concluding that premiums are rich, check whether the richness lives in one expiry rather than in the name. A stock can sit at the top of its annual range at the front month and look entirely ordinary further out, and selling the front-month premium because the rank is high means selling the very event that produced the rank. The second is comparative: the same trade can be cheap at one tenor and expensive at another, so the tenor is a decision rather than a default. A view about a slow re-rating belongs where the curve is not already charging for the event about to happen. The caveat belongs with the tool. A curve read from quoted implied volatilities is contaminated by the fact that strikes differ across expiries and by the liquidity differences between them, so a line drawn from whatever strike happens to be nearest the money is a rough instrument. What it is reliably good for is the question it was introduced to answer — how much of the premium in front of me is a scheduled event — and that question can be answered approximately and still change the trade. • A gently rising curve is normal; a steeply falling curve means an event sits inside the front expiry. • The gap between the front and the post-event expiry isolates the event’s contribution to premium. • A 65% front against a 30% back is roughly a twelve percent event move priced — a number to argue with. • High rank in the front month may be entirely event premium; sell the wrong tenor and the rank misleads. • The tenor is a choice: express a slow view where the curve is not already charging for a date. Compare expiries on the same name and, where possible, the same strike. A curve drawn from strikes at different levels of moneyness is partly a moneyness effect, which is the opposite of what you are trying to read.

Sources

• Implied volatility and expected moveHull, “Options, Futures, and Other Derivatives”
• IV rank and percentile methodologyPractitioner research on volatility-based position selection
• Volatility risk premium evidenceAcademic literature on the variance risk premium