Two comparisons with different denominators
Implied volatility (IV) is an annualized option-pricing input, not a forecast of returns or their direction. Cboe's VIX methodology describes an options-based measure of expected volatility for equities; our crypto pages use exchange mark IV and a constant-maturity ATM series, not the VIX calculation. Bybit's option ticker documentation distinguishes mark IV from bid and ask IV.
IV rank = (current IV − prior low) / (prior high − prior low) × 100. Rank locates the current reading in the prior range. A single extreme observation can stretch that range for the entire lookback.
IV percentile describes the share of prior observations below current IV. Tie conventions vary: here midrank percentile = (observations below current + 0.5 × observations equal to current) / reference count × 100. It measures the distribution of readings rather than distance between extremes. The current observation is excluded from both reference calculations.
Illustrative spike example: synthetic, not live data
Imagine 168 prior hourly observations spanning 7 days: 84 readings at 20% IV, 83 at 40%, and one spike at 100%. Current IV is 50%. These are synthetic annualized IV inputs, not market quotes.
The prior low is 20% and high is 100%, so rank = (50 − 20) / (100 − 20) × 100 = 37.5%. But 167 of the 168 prior readings are below 50%, with none equal: percentile = 167 / 168 × 100 ≈ 99.4%. Current IV is near the top of the observed distribution while still below the midpoint of the spike-expanded range. Neither result predicts what happens next.
What window this site uses
The IV rank tool uses the prior 30 calendar days of recorded hourly observations relative to the latest IV timestamp, not the common 52-week historical lookback. The underlying series is 30-day constant-maturity at-the-money IV: its forward maturity is separate from the backward-looking comparison window. Expiry-specific rows are excluded. Within each UTC hour only the last finite reading counts; gaps are not filled.
Coverage = unique reference hours / 720 × 100. There are 720 possible hours in 30 days. Span days is the elapsed time between the first and last reference samples, not a count of days with complete data. In the synthetic example, coverage is 168 / 720 × 100 ≈ 23.3%, despite a 7-day span. Read coverage, span and first/last timestamps together.
Both statistics require at least 168 prior hourly samples spanning 7 days. Current IV must be no older than 2 hours. Short history, stale readings, absent recorder history or restricted public venues leave comparisons unavailable; a dash is not zero. Only eligible public venue history contributes, and mixed venue coverage can change the comparison.
The constant flag means the prior low equals the high. Division by zero makes rank undefined; percentile still works, giving 50% when current IV equals every reference reading. The breakout flag means current IV is outside the prior range. Rank is not clamped: below 0% means a new low and above 100% a new high, while percentile remains between 0% and 100%.
How to use and misuse the comparison
“Cheap” or “expensive” here describes IV relative to its own observed history. Low rank does not establish that options are underpriced, and high percentile does not establish that selling them will profit. Future realized volatility, premiums, spreads and exposure to large moves still matter. A brief or gappy sample can describe a different regime from a full year.
Compare matching maturity, source, sampling and timestamps before comparing platforms or coins. Read implied volatility, smile and risk reversal for the construction of the series, and BTC versus ETH volatility for a comparison of the underlying IV measures.
Primary references checked on 17 October 2026: Cboe: VIX methodology · Bybit API: option ticker fields. Not financial advice.