Reading the VIX: What 20, 30 and 40+ Have Meant in Past Crises

Before the US open on 5 August 2024, the Cboe Volatility Index was above 65, a level previously reached only in the 2008 financial crisis and the March 2020 Covid crash. Front-month VIX futures stayed below 35. By the close the VIX was 38.57, and six sessions later it was back under 20. The S&P 500 fell 3.0% that day: a bad session, not a crash.

That morning is a compact lesson in reading the index. The headline number mixes the level of expected volatility, the shape of the curve across horizons, and the liquidity of the options behind it. This guide takes each in turn, then shows how CrisisBoard's published rules score a spike.

What the VIX measures

Cboe defines the VIX as a measure of the market's expectation of 30-day forward-looking volatility of US equities, as conveyed by S&P 500 option prices. It is not computed from stock prices. Cboe takes out-of-the-money S&P 500 puts and calls across a wide range of strikes from the two expiries that bracket 30 days, weights their mid-quotes in a variance-swap formula, and interpolates to a constant 30-day horizon. This method dates from 2003; the original 1993 design, based on at-the-money S&P 100 options, lived on as the VXO until September 2021.

The result is an annualized standard deviation in percentage points. To translate it into a daily move, divide by the square root of 252 trading days, about 15.9 (the "rule of 16"). For a 30-day move, divide by the square root of 12:

• VIX 15: about ±0.9% a day, ±4.3% over 30 days
• VIX 20: about ±1.3% a day, ±5.8% over 30 days
• VIX 30: about ±1.9% a day, ±8.7% over 30 days
• VIX 40: about ±2.5% a day, ±11.5% over 30 days
• VIX 80: about ±5.0% a day, ±23.1% over 30 days

These are one-standard-deviation ranges. If returns were normal and the VIX were an unbiased forecast, about one session in three would land outside them. In practice the VIX overstates the typical day. Across more than 9,200 sessions since 1990, the VIX averaged 19.4, while the S&P 500's realized volatility over the following 21 sessions averaged 15.4. The index exceeded the volatility that actually followed 85% of the time, and the S&P 500 moved by more than the prior close's implied daily range on only 18% of sessions. Option sellers charge a premium for carrying crash risk, and the VIX embeds it.

The tails make up the difference. On Friday 16 October 1987 the VXO closed at 36.37, implying a daily standard deviation of about 2.3%. The following Monday the S&P 500 fell 20.47%, close to a nine-standard-deviation day, and the VXO closed at 150.19.

The baseline: what 20, 30 and 40 mean

Cboe's daily history from 2 January 1990 to 19 October 2026 covers 9,297 sessions. The mean close is 19.4 and the median 17.6; the gap comes from a right-skewed distribution in which short spikes pull the average up. The 10th percentile is 12.2, the 90th is 28.5, and the record low close, 9.14, came on 3 November 2017. Bands, and what followed each, are more useful than any single threshold.

Three things stand out. First, a VIX in the 20s is ordinary: it covers 29% of all sessions, and three times in four no sharp sell-off followed. Second, risk rises non-linearly. A five-day S&P 500 decline of 5% or more, the moderate tier of CrisisBoard's US_market_drawdown rule, followed within a month about 4% of the time from a sub-15 reading, 25% of the time from the 20s and 43% from the 30s. Third, the 40-plus row is thin and concentrated: 160 of its 208 sessions fall in 2008, 2009 and 2020. Any statistic built on it mostly describes two crises.

Ten spikes, side by side

The peak alone is a poor guide. The two highest episode peaks, 80.86 in November 2008 and 82.69 in March 2020, came in genuine systemic crises, and in both the index stayed high. After Lehman Brothers failed on 15 September 2008, the VIX did not close below 20 again until 22 December 2009, and it closed at 40 or higher for 63 consecutive sessions from 2 October 2008. After the Covid peak, the first close below 20 came on 12 February 2021.

Positioning shocks look different. The VIX was back under 20 seven sessions after Volmageddon on 5 February 2018 and six sessions after 5 August 2024. The Bank for International Settlements found that the 20-point jump in February 2018, the largest daily rise since the 1987 crash, came largely from dynamics inside the volatility market, with the systematic rebalancing of leveraged and inverse volatility exchange-traded products a key factor. For August 2024, a later BIS bulletin pointed to dealers widening bid-ask spreads, particularly on puts, which lifted the mid-quotes the index is built from. An SEC staff working paper traced the pre-market surge mainly to higher mid-quotes on out-of-the-money puts.

In the flash crash of 6 May 2010, the SEC and CFTC staff report found, major US indices already down more than 4% fell a further 5% to 6% within minutes, then rebounded. The VIX touched 40.71 and closed at 32.80; its episode high came two weeks later, amid euro-area stress. The SVB run of March 2023 is the opposite case: a US bank run with a VIX closing high of 26.52. The index prices expected swings in the S&P 500, not stress in bank funding, which is why the SVB episode registered far more in regional bank shares and deposits than in the VIX.

Term structure: the shape tells you more than the level

Cboe publishes the same calculation at several horizons: VIX9D (9 days), VIX (30 days), VIX3M, VIX6M and VIX1Y. In calm markets the curve slopes upward (contango): the next few days look quiet and uncertainty accumulates with time. In acute stress it inverts into backwardation, pricing the coming days as more dangerous than the coming months.

Inversion is rarer and more specific than a high level. Since VIX3M data begin in September 2009, the VIX has closed above VIX3M on 7.5% of sessions: 1.2% of sessions when the VIX was below 20, 14% when it was between 20 and 30, and 56% when it was above 30. The 9-day measure is more sensitive. VIX9D has closed above the VIX on 26% of sessions since 2011, and on 67% of sessions when the VIX was above 30.

At the spikes in the table, the inversion was extreme. On 5 February 2018, VIX9D closed at 59.34 against a VIX of 37.32 and a VIX3M of 28.13. On 16 March 2020 the three readings were 106.66, 82.69 and 67.70. On 5 August 2024 they were 42.79, 38.57 and 33.71, and the VIX was back below VIX3M three sessions later.

In 2022, by contrast, the S&P 500 fell 25.4% from its January peak to its October low, and the VIX averaged 25.6. Yet it never closed above 36.45, and it sat above VIX3M on only 6% of sessions. A level in the mid-20s with an upward-sloping curve describes a market that expects to stay uncomfortable for months, not one that fears tomorrow. The level alone cannot tell those apart; the curve can.

Why a low VIX is not low risk

A low VIX means options are cheap, usually because realized volatility has been low. It says little about the odds of a jump; in several cycles it sat near its lows close to the market top:

• 2007.The VIX averaged 10.8 in the first eight weeks of 2007. On 27 February it rose 64% in a day to 18.31 as the S&P 500 fell 3.5%. It first closed above 30 on 15 August, and the S&P 500 made its pre-crisis peak on 9 October with the VIX at 16.12. The first close above 40 did not come until 29 September 2008.
• 2017 to 2018.The VIX averaged 11.1 in 2017, with 52 closes below 10. On 26 January 2018 the S&P 500 set a record with the VIX at 11.08. Six sessions later it closed at 37.32.
• 2020 and 2024.The S&P 500 peaked on 19 February 2020 with the VIX at 14.38, and on 16 July 2024 with the VIX at 13.19.

Calm can also build the conditions for its own end. Volatility-targeting strategies add exposure as realized volatility falls, and sellers of options and short-volatility products collect premium until a jump forces them to cover. The BIS called August 2024 "yet another example of volatility exacerbated by procyclical deleveraging and margin increases."

Beyond the US: VSTOXX and the Nikkei VI

Europe's equivalent, the VSTOXX (V2TX), applies a variance-swap method to Euro Stoxx 50 options listed on Eurex and interpolates to a constant 30 days. It runs structurally higher than the VIX: since 1999 its mean close is 23.1 and its median 21.0. Its record close, 87.51, came on 16 October 2008; it reached 85.62 on 16 March 2020.

The gap between the two indices hints at where a shock starts. During the euro crisis the VSTOXX peaked at 53.55 on 12 September 2011, above the VIX's 48.00. After Russia invaded Ukraine it closed at 49.64 on 4 March 2022, while the VIX's high for 2022 was 36.45. On 5 August 2024, a shock set off by weak US data and amplified by the unwinding of leveraged equity and currency trades, the VSTOXX closed at 31.17 against the VIX's 38.57.

Japan's Nikkei Stock Average Volatility Index is computed by Nikkei from Nikkei 225 futures and options on the Osaka Exchange, and has been published in real time since January 2012. Japan was at the centre of August 2024: the Nikkei 225 fell 12.4% on 5 August, a record one-day loss of 4,451 points. Each index has its own baseline, so a single cross-market threshold would misfire.

How CrisisBoard's volatility rules work

Each country and regional market on CrisisBoard has a volatility input defined in its published configuration. For the United States it is the VIX itself, read from the live quote. Elsewhere the input is realized volatility: the annualized standard deviation of log returns over the last 20 daily closes of a proxy, such as the Euro Stoxx 50 for Europe, the EWJ ETF for Japan, a FTSE 100 ETF for the UK and the EWZ ETF for Brazil. Two rules apply that input, with thresholds set by each market's normal turbulence:

• vol_elevated (US_vol_elevated, EU_vol_elevated and so on), 1 point: volatility above 25 in developed regions (US, Europe, UK, Japan, Australia, Asia ex-Japan) and above 35 in the emerging set (Argentina, Brazil, Turkey, Mexico, India, South Korea, Taiwan, China, the EM aggregate, the Gulf and Iran).
• vol_spike, 2 points: above 30 in developed regions, above 45 in the emerging set.

The US region adds two VIX-specific rules. US_vix_term_inversion scores 2 points whenever the 9-day VIX9D trades above the 30-day VIX, and 3 when it is two or more points above. US_vix_velocity_spike adds 2 points when the VIX jumps more than 15% within about five minutes, measured against its recent reading, so a sudden move from 15 to 22 registers before any level threshold does. Europe has a matching rule on its realized-volatility input, with a 12% threshold.

Added together, the US volatility rules can contribute at most 8 points, exactly the US MONITOR threshold. PHASE_1 starts at 10 and PHASE_2 at 12, as the methodology sets out. A volatility shock alone can make CrisisBoard watch more closely; it cannot declare a crisis. Escalation needs other families to move: regional banks (KRE), the S&P 500's five-day drawdown (SPY), high-yield and leveraged-loan credit (HYG, SRLN), T-bills (BIL), and software and BDC stress.

The realized-volatility inputs add a second brake: they lag, because 20 daily closes take time to absorb a shock. By our reconstruction, the Euro Stoxx 50's 20-day realized volatility peaked near 20 in early March 2018 and near 24 in mid-August 2024. EU_vol_elevated would not have fired in either episode, even though the VSTOXX closed at 31.17 on 5 August. Japan's EWJ-based measure, near 31 that day, tripped both JP_vol_elevated and JP_vol_spike.

The cross-region layers are stricter still. The Black Swan detector's VOL_STORM condition needs five or more regions above 35 (50 for emerging markets), and the detector activates only when at least two independent condition families align. On the regime page, the VIX weighs heavily in the sentiment score, but a risk-off call also needs average regional flows below −15, and a dislocation call needs four or more regions at PHASE_1 or worse.

The replay below includes both T-bill rules. US_funding_stress fires when BIL's total return, dividends added back, falls 0.08% in a day or 0.15% over five days. US_flight_to_safety adds 4 points when BIL's five-day total return is positive and the day's volume is at least twice its 20-day average.

Five questions for any VIX spike

1. Where is the close in the long-run distribution? A close above 30 puts the index in roughly the top 8% of sessions since 1990; above 40, the top 2%.
2. Is it a close or an intraday print? Pre-market and intraday highs are built from quotes and can overshoot.
3. What is the curve doing? VIX9D above VIX, and VIX above VIX3M, mark acute stress. A high level with an upward-sloping curve marks persistent uncertainty.
4. How long does it last? The 2018 and 2024 shocks were under 20 within seven sessions. Systemic crises held the VIX above 40 for weeks.
5. Do other markets confirm? Check credit, bank shares, funding, non-US volatility, and which CrisisBoard signal families are firing.

The VIX is the fastest stress gauge in finance and one of the least discriminating. Alone, it flags positioning accidents and systemic crises with similar numbers; with its curve, its duration and confirmation from other markets, it tells them apart.