Skip to main content

VIX & Volatility Gauges

Updated Jun 24, 2026 at 2:35pm

Research Draft Medium 1,196 words

The VIX is the Cboe Volatility Index — the market's option-implied estimate of how much the S&P 500 will move over the next 30 days, expressed as an annualized percentage. It is widely called the "fear gauge" because it tends to spike when stocks fall hard. The core tension is that VIX is genuinely informative about current option demand and crisis stress, yet it is a notoriously poor forecaster of future realized volatility — it tells you what the option market is charging for protection today far more reliably than what volatility will actually be tomorrow. A whole family of related gauges (VIX9D, VIX3M, VVIX, MOVE) extends the same idea across horizons, across the volatility of volatility, and into the bond market.

How it's calculated / formed

VIX is not the implied volatility of a single option. Since the 2003 redesign (Cboe with Goldman Sachs), it is a model-free variance estimate: a weighted strip of out-of-the-money SPX put and call prices across all available strikes, which approximates the price of a 30-day variance swap, then converted to an annualized volatility. Inputs are SPX and SPXW (weekly) option mid-prices plus Treasury yields. The calculation selects near-term options (closest expiry under ~30 days) and next-term options (closest over 30 days) — Cboe uses options in roughly the 23–37 day window — computes the variance of each, and linearly interpolates to a constant 30-day figure (Cboe methodology; Wikipedia). A VIX of 20 means the option market is pricing an annualized standard deviation of ~20%, i.e. roughly a 5.8% monthly move (20 ÷ √12).

Cboe publishes a constant-maturity term structure: VIX9D (9-day), VIX (30-day), VIX3M (3-month), VIX6M, and VIX1Y. VVIX measures the implied volatility of VIX options — the "volatility of volatility," or the acceleration of fear versus VIX's velocity (Wikipedia). The MOVE index (Merrill/ICE) is the bond-market analog, derived from Treasury option implied vol. Crucially, VIX itself cannot be traded directly; exposure comes via VIX futures, options, and ETPs (VXX, UVXY, SVXY), which track futures — not spot — and therefore decay or diverge, sometimes badly.

How it's used in practice

Three distinct uses, roughly in order of how well they hold up:

1. Stress/regime reading (strongest). Crisis levels are unambiguous: VIX closed at ~80.86 on 20 Nov 2008, with the all-time intraday high of ~89.53 set 24 Oct 2008; it then set a record close of ~82.69 on 16 March 2020 (Macroption; Wikipedia). "Normal" sits roughly 12–20. A VIX above ~30 reliably signals that the market is paying up steeply for protection right now.

2. Term-structure / curve shape. VIX futures trade in contango (upward-sloping, calm) roughly 80% of the time and flip to backwardation (downward-sloping, acute stress) the remaining ~20% (Volatility Box; growyourpile). A practical proxy is the VIX/VIX3M ratio: below 1.0 is contango, above 1.0 signals the front end has inverted — a real-time stress flag. Contango also drives the persistent negative roll yield that erodes long-VIX ETPs.

3. Contrarian sentiment overlay. Because high VIX coincides with panic and depressed prices, it is used as a mean-reversion tell: extreme spikes often mark capitulation lows. Research has found an optimal high-VIX threshold around the 80th–85th percentile as a useful explanatory term for subsequent excess equity returns (Advisor Perspectives, summarizing academic work). Low VIX is loosely read as "complacency," but this signal is far weaker — VIX can stay low for long stretches.

Adoption, debate & evidence

VIX is one of the most-watched numbers in finance and the underpinning of a large listed-derivatives complex — its institutional adoption is not in dispute. What is contested is what it predicts.

  • Folklore: "VIX forecasts the next month's volatility." Measured: model-free implied vol forecasts realized vol only modestly better than naive backward-looking measures; researchers note the predictive power of volatility models is often "similar to plain-vanilla measures such as simple past volatility" (Wikipedia, citing the forecasting literature). Robert Shiller showed a retrospectively computed 1929 VIX failed to anticipate the Depression's realized volatility.
  • The robust empirical fact is the bias, not the forecast. The volatility risk premium — VIX minus subsequently realized 30-day vol — has been positive roughly 86% of the time, averaging about 4.2 vol points since 1990 (Barclays). This is why short-volatility strategies earn a premium and why VIX systematically overstates future volatility. The premium is widely attributed to the negative correlation between returns and volatility (investors pay up for crash insurance).
  • Sharp caveat: that same short-vol premium hides ruinous tail risk. "Volmageddon" (5 Feb 2018) saw VIX jump ~115% in a day (close ~17.31 → ~37.32) — its largest one-day percentage move — wiping out inverse-VIX products like XIV (which collapsed from ~$1.9B to ~$63M in a session and was terminated) overnight (sixfigureinvesting; Wikipedia). The premium is "picking up pennies in front of a steamroller."

Strengths & limitations

Works best as a coincident stress and capitulation gauge, and as the input to harvesting the volatility risk premium. It is forward-looking, model-free, real-time, and hard to manipulate at the index level.

Fails as a precise volatility forecast and as a market-timing tool on the low side — a low VIX is not a sell signal and complacency can persist. The #1 misuse is treating VIX ETPs (VXX, UVXY) as a clean proxy for spot VIX: they track futures, bleed via contango roll, and decouple from the index, so "VIX is at 15, that's cheap insurance" reasoning via VXX is a classic, expensive error. A secondary misuse is reading any single absolute level without the term structure — VIX 25 in steep backwardation means something very different from VIX 25 in contango.

Sources

Disputes flagged: (1) VIX's forecasting power for realized volatility is genuinely contested — strong "fear gauge" framing overstates it. (2) Some statistics (80% contango, 86%/4.2-point VRP) come from practitioner sources whose exact sample windows vary; treat as commonly cited orders of magnitude, not precise constants.