What it means
A volatility cone plots realised volatility for an FX pair across multiple lookback windows (10-day, 30-day, 60-day, 120-day, 252-day) showing percentile bands (10th, 25th, 50th, 75th, 90th) over a long history. Current realised vol is overlaid against the cone to show where today sits relative to the pair's historical regime. A 30-day realised vol below the 10th percentile of the historical distribution signals exceptionally calm conditions — often the setup for a regime break.
Why it matters
Vol cones are the cleanest way to spot mean-reversion setups in FX vol. When a pair's 30-day realised vol compresses below the 10th percentile of its historical distribution for 3+ weeks, the probability of a vol-expansion event rises sharply. The August 2024 USD/JPY unwind was preceded by 30-day realised vol below 6% — historic-floor territory for the pair.
How to use it
Build the cone in Python or trading platform: pull 5+ years of daily returns, compute rolling realised volatility for windows 10/30/60/120/252, plot percentile bands. Overlay current realised vol. Trade signals: vol < 10th percentile = compressed regime, vol-expansion bets favoured; vol > 90th percentile = exhausted regime, vol-contraction bets favoured.
USD/JPY 30-day realised vol on 25 July 2024: 5.8%, 7th percentile of 10-year history. Implied vol on 1-month ATM options at the time: 7.2%, also low decile. Pure vol-expansion setup. By 5 August 2024, realised vol had spiked above 18% — 99th percentile reading — as the carry unwind cascaded.
Building a vol cone — the data and math
Inputs: 5+ years of daily close-to-close FX rates for the target pair. Compute daily log returns: r_t = ln(P_t / P_{t-1}). Compute rolling annualised volatility for windows 10, 30, 60, 120, 252: σ_window = std(r_t over window) × √252. Plot the percentile bands (10/25/50/75/90) of each window's distribution. Overlay current realised vol point for each window. The visual: a funnel-shaped cone (wide at short windows, narrow at long) with current vol marked.
Reading the cone — three useful signals
(1) Compressed regime: current vol below 25th percentile for all windows = vol-expansion bias. The pair has 'used up' its mean-reverting low; reversion to median is the path of least resistance. (2) Expanded regime: current vol above 75th percentile for all windows = vol-contraction bias. The pair has overextended; revert-to-median is now the play. (3) Term-structure tilt: short-window vol > long-window vol = recent shock; short-window < long-window = recent calm; the steepness tells you the dominant regime.
- Below 10th percentile: extreme compression, vol-expansion probability >60%
- 10-25th percentile: compressed, vol-expansion lean
- 25-75th percentile: normal regime, no edge
- 75-90th percentile: expanded, vol-contraction lean
- Above 90th percentile: extreme expansion, vol-contraction probability >60%
Realised vol vs implied vol — the volatility risk premium
Vol cones use realised volatility (computed from historical price moves). Implied volatility comes from option prices. The spread between them is the volatility risk premium (VRP). Historically, implied vol > realised vol by ~1-3 vol points on majors — option sellers earn this premium for taking the risk of vol expansion. When implied vol approaches or falls below realised vol, the VRP has collapsed, signaling complacency. The combination — realised vol in the bottom decile AND VRP collapsed — is the highest-conviction vol-expansion setup in FX.
Vol regime examples across major pairs
Historical low/high realised vol for major pairs (10-year window): EUR/USD: low 4%, high 18%. GBP/USD: low 5%, high 22%. USD/JPY: low 5%, high 20%. AUD/USD: low 6%, high 30%. USD/MXN: low 8%, high 45%. JPY crosses (GBP/JPY, EUR/JPY) and EM pairs sustain higher base volatility and reach much higher peaks during crises. Each pair has its own characteristic floor and ceiling.
Trading vol cones — three structures
(1) Long-vol bet during compression: buy an ATM straddle (or option strangle for cheaper) with 30-60 day expiry. Profits if realised vol exceeds implied vol over the holding period. (2) Short-vol bet during expansion: sell straddles with tight risk control, or sell variance swaps to institutional. (3) Calendar spread: sell short-dated, buy long-dated — profits if the short-end vol normalises faster than long-end. Each has different Greeks exposure.
Limitations of vol cones
Cones assume stationary distributions — that the future will resemble the past. FX regimes can shift structurally (e.g., post-2022 dollar strength regime). A 10-year cone in 2026 includes the 2020 pandemic, 2008 GFC and 2015 SNB break — fat-tail events that distort the historical distribution. Adjust the lookback window to recent regime. Cones are pattern signals, not prophecy: a compressed vol regime can stay compressed for 6+ months before a break.
Frequently asked
What is the difference between realised and implied volatility?
Realised volatility is computed from actual historical price moves (standard deviation of log returns, annualised). Implied volatility is the volatility figure that makes the Black-Scholes formula match observed option prices — the market's forecast of future vol. RV is backward-looking, IV is forward-looking.
Why do volatility cones funnel inward?
Longer-window realised volatility is calculated over more data points, smoothing out short-term noise. A 252-day vol is less variable across history than a 10-day vol. Visually, the percentile bands are tight for long windows and wide for short windows — the cone shape.
How often should I rebuild a vol cone?
Recompute weekly during stable regimes, daily during transitions. The percentile bands are computed once over a long history and updated quarterly. The 'current vol' point moves with each new data point.
Can I use vol cones for non-FX assets?
Yes. The math is asset-agnostic. Equity index futures, commodity futures, single stocks, and crypto all support volatility cone analysis. The percentile thresholds and base vol levels differ by asset class.
What signals a vol-expansion trade?
Three conditions together: current realised vol in the bottom decile of historical distribution, implied vol approaching or below realised vol (compressed risk premium), and a near-term scheduled catalyst (FOMC, BoJ, NFP) within the option's expiry window. When all three align, the long-vol bet has highest expected value.
Is vol mean-reverting?
Yes, on multi-month timeframes. Realised vol exhibits mean-reverting behaviour: compressed regimes give way to expansion, expansion gives way to compression. The half-life of vol mean-reversion in majors is roughly 30-90 days. Short-term, vol can persist in regime for weeks before reverting.
Want a worked example or a deeper dive? Ask Rocky how this concept applies to your specific watchlist or trade idea.
Ask Rocky