Vol regime & term structure

How to tell when options are cheap, when they’re expensive, and when the market is pricing tail risk.

These are the tools for reading what the options market is saying — separate from what the cash equity market is doing. Often the two disagree, and that’s where the signal is.

The vol regime verdict

The Volatility panel on Equities opens with a verdict, not a score — five states derived from explicit rules, in strict priority order:

State	What it means
Panic	The VIX curve is inverted (backwardation) — or the fast leg is: the 9-day trading over the 30-day counts even while VIX/VIX3M sits in contango, and the read says which one fired. A vol event is in progress — hedges are paying now; don’t sell vol into it.
Underpriced risk	Realized vol is outrunning implied. The market is under-pricing delivered risk — the cheapest honest moment to hedge.
Rich premium	Options are paying sellers well above delivered vol. Protection is overpriced — favor covered calls or trimming hedges over buying puts.
Cheap protection	Low VIX and thin premium — insurance is genuinely cheap, not just optically low.
Normal	Vol is priced about right for what the tape is delivering.

The spine of the verdict is the IV−RV premium: VIX minus 20-day realized S&P volatility — the actual price of protection. A low VIX is not the same as cheap options (2017: VIX 11 vs realized 6 = expensive), and a high VIX is not the same as expensive ones (2020: VIX 30 vs realized 60 = cheap). Comparing implied to delivered is what separates a pricing read from a fear gauge.

Under the verdict sit six dimension rows — premium, term structure, VIX level, implied correlation, dispersion, vol-of-vol — each with its percentile strip and sparkline. There’s deliberately no composite 0–100 score: any weighting of those dimensions would be arbitrary, so the panel states its rules and shows its inputs instead.

The one-line read follows the same shape as every verdict panel on the site: the state, then the evidence for it, then an outlier only when one clears a threshold — if nothing qualifies, nothing is appended. The gated outliers here are the VVIX early warning (below) and the fragile-calm flag from implied correlation and dispersion — and since those last two are one phenomenon in two framings, when both are at their extremes they collapse into a single combined sentence rather than repeating themselves.

VIX term structure

VIX (1-month implied vol) divided by VIX3M (3-month).

• Ratio > 1 — backwardation. Near-term fear exceeds longer-dated — traders are paying up for immediate protection. The structural panic signal; it overrides everything else in the verdict.
• Ratio < 1 — contango. Normal — longer-dated vol is higher than spot, the market expects calm to persist.
• Rising toward 1 = stress building; falling deeper below 1 = calm reasserting.

Both legs of every ratio come from the same session — a stale leg can never fabricate an inversion.

Implied correlation (COR3M)

CBOE’s 3-month implied correlation index — why index vol is where it is. Index variance is roughly average single-stock variance times average correlation, so when VIX is low because correlation is priced near record lows, the calm is dispersion-driven and fragile: a correlation spike lifts index vol mechanically, even if no single stock gets more volatile.

The row’s percentile ranks against the index’s full ~20-year history (these extremes are generational — a 1-year window would hide them). It never drives the verdict; it colors it. At a bottom-decile extreme the read appends the fragile-calm outlier: “Index calm rests on p1 implied correlation — dispersion-priced; a correlation spike lifts index vol mechanically.”

Dispersion (VIXEQ − VIX)

The same story in vol points: Cboe’s equal-weight single-stock implied vol index (VIXEQ) minus the VIX. It answers “how much wilder are individual stocks than the index?” — which is exactly what a low implied correlation buys. The row’s percentile ranks against the full VIXEQ history (Cboe’s backcast reaches 2014), because dispersion extremes, like correlation extremes, are generational.

Where implied correlation tells you the assumption (stocks will offset), dispersion tells you the magnitude riding on it — at record readings, single-stock vol can run 30+ points over the index, meaning index calm rests entirely on winners and losers continuing to cancel. Like implied correlation, it colors the verdict but never votes: they’re two framings of one phenomenon, and one phenomenon gets one vote.

The row is two-tailed, and the tails mean different things. A record-wide spread flags amber — fragile calm, a conditional warning: nothing has broken, but the index’s stillness depends entirely on correlation staying pinned. A compressed spread flags red — stocks and index moving as one is the correlation-stress signature, what crashes actually look like. Mirror-image of the implied-correlation row’s coloring, so the two framings always agree.

SKEW

A separate gauge: how much OTM puts cost relative to ATM options. Tracked as a percentile.

• High SKEW (p90+) = market is paying up for tail-risk protection. Either a real worry or a smart-money hedge.
• Low SKEW (p10–) = complacency about the left tail.

SKEW is shown in the volatility table but carries no weight in the verdict — it’s constructed from thinly-traded deep-OTM quotes and is too noisy to trust as a signal on its own. Read it as color, not conviction.

Vol of vol (VVIX)

The implied volatility of VIX options. A VVIX spike while spot vol sleeps — someone bidding for VIX calls into a calm tape — is the classic early warning. When VVIX runs 2σ above its yearly norm it appends a ⚠ chip and a warning clause to the read — except during Panic, where an elevated VVIX is expected rather than an outlier.

Where you see it

• The Volatility panel on Equities — the verdict lives here
• VRP Vol Harvesting model — the deep-dive on the same premium, with harvest/hibernate regimes and strategy simulations
• Cross-Asset Verdict section of the Daily Recap
• The Market Pulse briefing reads the verdict directly