What Is the VIX? The Market's Fear Gauge, in Real Numbers

Michael Whitman · Trading since 2017, full-time since 2020 · Published 2026-09-25

The VIX is the number people mean when they say “fear is up.” It is worth knowing exactly what it measures, because it is not what most headlines imply.

How it works

Cboe describes the VIX as “a leading measure of market expectations of near-term volatility conveyed by S&P 500 Index (SPX) option prices” (cboe.com, VIX product page, read 25 Sep 2026). It was introduced in 1993, and VIX futures followed in 2004.

Built from S&P 500 option prices

It is built from options, not from stock prices. When traders pay more for S&P 500 options, especially the puts that protect against a fall, the VIX rises. When option prices are cheap, it falls. The index tracks what the market is paying for protection, which is why it gets called the fear gauge.

It always looks 30 days ahead. Cboe’s own FAQ explains that the options behind it change slightly every minute “in order to maintain a constant maturity of 30 days” (cboe.com, VIX FAQs, read 25 Sep 2026).

You cannot buy the VIX itself. The same FAQ says traders cannot buy and hold the options behind the index, because the mix would need rebalancing all the time. Products that track it use futures, and futures behave differently from the spot number on your screen.

The number is stated as a yearly percentage. A VIX of 20 means the options are priced for moves of about 20% a year, one standard deviation, in either direction. That is the source of the “rule of 16” used below: a year has about 252 trading days, and the square root of 252 is close to 16.

It moves opposite to stocks most of the time. Falling markets make people buy protection, so the VIX tends to jump when the S&P 500 drops and drift lower when it climbs. It is a reading on the whole index, the same way implied volatility is a reading on one option.

A worked example

Take a real close: 18 Sep 2026. The VIX closed at 14.81 and the S&P 500 closed at 7,650.50, both from the daily files described below.

Step one, the daily figure. Divide the VIX by the square root of 252, about 15.87. 14.81 divided by 15.87 is 0.93%. That is the size of a one-standard-deviation day the options were priced for.

Step two, turn it into points. 0.93% of 7,650.50 is about 71 points. On that evening, option prices implied that the next day’s S&P 500 move would land inside roughly 71 points either way about two days in three.

Step three, the monthly figure. Divide by the square root of 12 instead: 14.81 divided by 3.46 is 4.28% for the coming month.

None of this is a prediction of direction. The VIX says how far, not which way. A 71-point day up and a 71-point day down fit the same reading equally well.

The original data

The data: every daily close of the VIX from 2 Jan 1990 to 24 Sep 2026, 9,252 trading days, downloaded from Yahoo Finance (^VIX) on 25 Sep 2026. The full file is published as a CSV of every VIX close so the counts below can be checked.

The median close was 17.58. The lowest was 9.14 on 3 Nov 2017; the highest was 82.69 on 16 Mar 2020, with 80.86 on 20 Nov 2008 close behind. The VIX closed under 20 on 5,824 days, 62.9% of the total, and at 30 or more on 736 days.

By range: under 12 on 801 days (8.7%), 12 to 15 on 2,159 (23.3%), 15 to 20 on 2,864 (31.0%), 20 to 25 on 1,827 (19.7%), 25 to 30 on 865 (9.3%), 30 to 40 on 528 (5.7%), and 40 or more on 208 (2.2%).

The rule of 16 against real closes

Then the rule of 16, tested. Matching each VIX close to the next day’s S&P 500 close gives 9,245 pairs. On 7,548 of them, 81.6%, the S&P 500’s actual move was smaller than the VIX-implied one-day figure. On the 736 days the VIX closed at 30 or higher, that share fell to 72.4%. Options usually priced in more movement than arrived, which is the volatility risk premium showing up in real data.

And spikes take a while to fade. The VIX has closed at 40 or more in nine separate episodes since 1990. Counting from the first such close to the first close back under 20, the median wait was 114 trading days. The shortest was 25 trading days, in April and May 2025. The longest was 311, from September 2008 to December 2009.

In the 24,971-video corpus this site studies, 15 have VIX in the title, from 10 channels, at a median of 8,626 views. Several of those are about a different instrument or an indicator that borrows the name, so the gauge itself is thinly explained.

When it fails

The first failure is reading it as a direction call. A high VIX says big moves are priced in, not that stocks will fall. The ten largest one-day S&P 500 gains since 1990 all came the day after a VIX close above 45, the biggest a rise of 11.6% on 13 Oct 2008.

The second is treating a low VIX as safety. The lowest close in the file, 9.14 on 3 Nov 2017, was followed three months later by a close of 37.32 on 5 Feb 2018. A calm reading is the price of protection today, not a promise about next month.

A third is assuming the rule of 16 is a boundary. The next-day move was smaller than the implied figure 81.6% of the time, which also means it was larger on 1,697 of the 9,245 days. Those are the days that decide whether a position survives.

A fourth is expecting a spike to reverse quickly. The median wait from 40 back under 20 was 114 trading days, more than five months. Selling volatility because “it always comes back down” is right about the direction and often wrong about the timing.

And a fifth is trading products that track it as if they were the index. Cboe’s own FAQ says you cannot hold the VIX. Funds and notes that follow it use futures, which can lose value steadily even when the spot VIX is flat.

Related

Implied volatility explains the same measurement for a single option, which is where the VIX’s inputs come from. The volatility risk premium covers why options tend to price in more movement than arrives, as the 81.6% figure above shows. And options sets out the contracts themselves, if the idea of paying for protection is new.

This page is educational, not financial advice. Test every idea on your own charts before risking money.