Volatility Term Structure and Events: Contango, Inversion, Event Vol, and FOMC and CPI Weeklies

The volatility term structure is the pattern of implied volatility across expiries: the volatility the market prices for the next week, the next month, the next year. In calm markets short-dated implied volatility usually sits below long-dated (an upward slope, often called contango), and in stress it jumps above (an inverted or backwardated curve). Scheduled events such as Federal Reserve decisions, inflation reports, and earnings put bumps in the curve at the expiries that contain them.

In short: implied volatility is not one number per stock or index but a curve, and each point on it is an average of the variance expected between now and that expiry. Because variance adds up over time, two neighbouring expiries let you back out the volatility the market expects between them (forward volatility), and an expiry that contains an event lets you back out the volatility priced for the event day itself. Professionals trade the shape of that curve, often with calendar spreads. In Cboe’s own index data, the curve has been inverted on about 8% of trading days since 2009, and short-dated S&P 500 volatility rises measurably into Fed decision days and falls right after them.

This piece builds on implied volatility explained and the strike dimension covered in volatility surface and skew. It explains mechanisms and evidence. It is not trading advice.

A common misconception: “the stock’s implied volatility is 25%”

An options screen often shows a single implied volatility for a stock, but that figure is a choice, usually 30-day at-the-money. A one-week option and a one-year option on the same stock can carry very different implied volatilities, and the difference carries information: about events the market expects, about how quickly it thinks current turbulence will fade, and about the price of insurance at different horizons. Comparing a 10-day option’s implied volatility to a 90-day option’s without accounting for what lies inside each window is a common way to misread “cheap” and “expensive.”

Term structure shapes

Three shapes cover most of what you will see:

Upward sloping (contango). Short-dated implied volatility below long-dated. The normal state: current realized volatility is low, and longer horizons carry more room for something to happen, plus a premium for uncertainty.
Inverted (backwardation). Short-dated above long-dated. The market is pricing turbulence now that it expects to fade.
Humped. A bump at one or two expiries, typically because they contain a scheduled event such as earnings, a central bank decision, or an election.

What the data show

Cboe publishes constant-maturity S&P 500 volatility indexes at several horizons: the VIX (30 days), VIX9D (9 days), VIX3M (3 months), and VIX6M (6 months). From Cboe’s daily history files (our calculation, September 18, 2009 to November 9, 2026, 4,312 trading days):

The VIX closed above VIX3M (an inverted 1-to-3-month curve) on 7.6% of days.
The median ratio of VIX to VIX3M was 0.884: in a typical market, 30-day implied volatility sits about 12% below 3-month.
Inversions cluster in stress years: about 20% of days in 2011, 21% in 2018, 21% in 2020, and 11.6% in 2025, against 0% in 2023 and under 1% in 2012, 2017, and 2021.

The steepest inversions in the data are a list of market shocks:

Date	VIX	VIX9D	VIX3M	VIX6M	VIX / VIX3M
August 24, 2015	40.74	54.61	31.14	27.36	1.31
February 5, 2018	37.32	59.34	28.13	24.55	1.33
March 16, 2020	82.69	106.66	67.70	55.40	1.22
April 8, 2025	52.33	67.63	41.50	35.77	1.26
November 9, 2026	19.92	19.88	20.85	22.22	0.96

The highest single ratio in the series was 1.34 on February 28, 2020. On each stress date, the shorter the horizon, the higher the volatility: the market expected the shock to fade, and priced the next nine days far above the next six months. The last row, as of the most recent close in the file, is the ordinary upward slope.

Why the slope depends on the level

Volatility tends to mean revert: turbulent periods calm down and calm periods do not last forever. Longer-dated options average over more of that reversion, so their implied volatility moves less than short-dated implied volatility and stays closer to a long-run level. When current volatility is low, the curve slopes up toward that level; when it is high, the curve slopes down. Splitting the same Cboe data by the level of the VIX shows it clearly:

VIX level	Trading days	Median VIX / VIX3M	Share of days inverted
Below 15	1,454	0.849	0.0%
15 to 20	1,573	0.884	2.3%
20 to 30	1,028	0.927	14.1%
30 and above	257	1.011	56.4%

In the data, the curve never inverted with the VIX below 15, and it was inverted on more than half of the days with the VIX at 30 or above. A trader who reads the term structure without the level is missing half the story: an inverted curve at a VIX of 35 is ordinary for a stressed market, while a flat curve at a VIX of 14 is unusual and worth explaining, for example by a scheduled event inside the 30-day window.

The VIX index itself, and the futures curve built on it, are covered in the VIX in depth.

Forward volatility: what the curve implies between expiries

Implied variance (volatility squared) times time is total implied variance, and total variance adds across non-overlapping periods. So for two expiries T1 < T2 with implied volatilities σ1 and σ2:

Forward variance (T1 to T2) = (σ2² x T2 - σ1² x T1) / (T2 - T1)

Worked example: forward vol from two expiries

A 30-day option at 18% and a 60-day option at 20%:

Total variance to 30 days: 0.18² x 30/365 = 0.002663
Total variance to 60 days: 0.20² x 60/365 = 0.006575
Forward variance: (0.006575 - 0.002663) / (30/365) = 0.0476
Forward volatility, days 30 to 60: √0.0476 = 21.8%

The second month is priced at 21.8%, above both quoted volatilities: a modest upward slope in spot volatilities implies a steeper slope in forward volatilities. Now invert the curve, 30-day at 30% and 60-day at 25%: forward volatility for days 30 to 60 is 18.7%. The market is pricing a turbulent first month followed by a much calmer second.

Forward variance cannot be negative. If σ2² x T2 were less than σ1² x T1, a trader could buy the longer option and sell the shorter one for a credit in total variance terms, which is a calendar arbitrage. Market makers’ surface-fitting software checks for it. This constraint is the reason a later expiry can carry lower implied volatility only by a limited amount.

Event volatility

The same additivity lets you isolate a single day. Price the days without an event at a base volatility, read off an expiry that does not contain the event, and attribute the extra variance in the expiry that does contain it to the event day.

Desks do this in trading-day time (252 days a year) rather than calendar time, because weekends and holidays carry little variance. Using calendar days would make the Monday after a weekend look artificially expensive.

Worked example: extracting FOMC-day volatility

A hypothetical index trades at 6,000 at Monday’s close. A Federal Open Market Committee decision is due Wednesday at 2:00 p.m. Eastern. Two daily expiries are quoted:

Tuesday expiry, 1 trading day, no event: 13% implied volatility.
Wednesday expiry, 2 trading days, includes the decision: 19% implied volatility.

Treat Tuesday as a base day at 13%:

Total variance to Wednesday (in annualized-vol-squared trading days): 0.19² x 2 = 0.0722
Tuesday’s share: 0.13² x 1 = 0.0169
Event day variance: 0.0722 - 0.0169 = 0.0553
Event day volatility: √0.0553 = 23.5% annualized

Converted to a one-day move, divide by √252: the event day is priced at a standard deviation of 1.48%, against 0.82% for a normal day. For a normal distribution the expected absolute move is the standard deviation times √(2/π), so the market is pricing an average move of about 1.18% on decision day (about 71 index points) versus 0.65% (about 39 points) on an ordinary day. The event accounts for 77% of the variance in the Wednesday option.

The same arithmetic is how earnings moves are read from single-stock options; earnings and the implied move works through that version and the evidence on whether it is accurate.

FOMC and CPI weeklies

The listings that made this possible

Short-dated index options now exist for every trading day. Per Cboe’s April 13, 2022 announcement, SPX Weeklys, which already had Monday, Wednesday, and Friday expirations, added Tuesday expirations from April 18, 2022 and Thursday expirations from May 11, 2022. Since then a trader can choose an expiry that ends just before or just after a scheduled release. SPX options are European and cash-settled, so there is no early exercise to complicate the event pricing.

Trading has concentrated in the shortest dates. Cboe’s full-year 2025 volume report (January 6, 2026) said SPX options averaged 3.9 million contracts a day in 2025, and SPX zero-days-to-expiry options a record 2.3 million, 59% of SPX volume. Its July 2026 report (August 5, 2026) put 0DTE at a record 66.2% of SPX volume for that month. These figures change monthly; they are as of those reports. Same-day options are covered in zero DTE options explained.

What the curve does around Fed decisions

To see event pricing in public data, we compared VIX9D (the 9-day index) with the VIX (30 days) around the 47 scheduled FOMC decision days from January 2021 to October 2026, using Fed meeting dates from the Federal Reserve’s FOMC calendar and Cboe’s daily closes:

VIX9D / VIX ratio	Median
All trading days, 2021 to November 2026	0.928
Close on the day before an FOMC decision	1.000
Close on the day after an FOMC decision	0.939

Normally, 9-day S&P 500 volatility trades about 7% below 30-day. On the eve of a Fed decision, with the event inside a much shorter window, the ratio rises to about 1. Two days later it is back near normal. Measured directly, VIX9D fell from the close before the decision to the close after it on 72% of FOMC dates, with a median fall of about 10%. Across all two-day windows since 2021, it fell 53% of the time with a median change of about -1%. This is event volatility crush at the index level: the uncertainty premium is paid out or released once the decision is known.

We ran the same check on 56 consumer price index releases from February 2022 to October 2026, using release dates from the Bureau of Labor Statistics archive. CPI is published at 8:30 a.m. Eastern, before the stock market opens, so the before-and-after comparison is the prior close against the release-day close. The median VIX9D / VIX ratio was 0.965 on the evening before a release and 0.915 at the close of release day. The bump is smaller than for FOMC days, but the pattern is the same.

These are our calculations from public index closes. Event windows overlap (19 of the FOMC decisions came within nine days of a CPI release), VIX9D is a 9-day average rather than a single-day measure, and the sample is a few years long, so treat the numbers as a description of the pattern, not as constants.

Calendar spreads as volatility trades

A calendar spread sells a shorter expiry and buys a longer one at the same strike. Its basic payoff and Greeks are covered in the strategies section article on calendar spreads. As a volatility trade it is a position in the term structure: short the front’s gamma and the front’s event variance, long the back’s vega.

Worked example: a calendar across an FOMC decision

Same hypothetical index at 6,000, Monday close, decision Wednesday. Assume the market prices every non-event day at 13% and the event day at the 23.5% extracted above. Then consistent implied volatilities are:

Front: this Friday’s expiry, 4 trading days including the decision: 16.3%
Back: next Friday’s expiry, 9 trading days including the decision: 14.6%

The front’s volatility is higher because the event is a larger share of a shorter window. A trader buys the at-the-money next-Friday straddle for 131.62 and sells this Friday’s at-the-money straddle for 98.18, a debit of 33.44 points ($3,344 with a $100 multiplier). The front straddle has a vega of about 6.0 points per volatility point and the back about 9.1.

At Wednesday’s close, after the decision, assume both expiries revert to 13% and the index has moved once, on the announcement:

Index move on decision day	Front straddle (2 days left)	Back straddle (7 days left)	Spread value	P&L (points)	P&L at $100 multiplier
0.00%	55.44	103.72	48.28	+14.84	+$1,484
0.50%	60.66	106.73	46.07	+12.63	+$1,263
1.00%	75.11	115.04	39.94	+6.50	+$650
1.25%	85.13	121.07	35.94	+2.51	+$251
1.50%	96.57	128.25	31.67	-1.77	-$177
2.00%	122.51	145.70	23.20	-10.24	-$1,024
3.00%	180.24	190.46	10.22	-23.22	-$2,322

The breakeven is a move of about 1.40% in either direction. The trade profits if the decision-day move comes in below that and loses beyond it. Notice what the trade actually is: even though the back option has more vega, both legs contain the same event, so the “IV crush” mostly hits the front, whose value was more concentrated in the event. The P&L is driven by the front’s short gamma. A calendar like this is a bet that the event move will be smaller than priced, not a pure bet on implied volatility falling. A calendar whose front expiry ends before the event and whose back contains it is the opposite trade: long the event.

Other term structure trades

Long-dated vega against short-dated gamma. Long-dated options have more vega and less gamma; short-dated the reverse. A book that is vega flat can still be heavily short or long gamma.
Weighting vega by maturity. Long-dated implied volatility moves less than short-dated, roughly in proportion to one over the square root of time in many desks’ rules of thumb, so risk systems often “time-weight” vega before adding it up across expiries.
Event strips. Some desks build a calendar of event days (FOMC, CPI, payrolls, earnings) and price each expiry as base variance plus the sum of event variances inside it, then trade expiries that look out of line.

What can go wrong

The event is bigger than priced. Short front calendars and short event straddles lose quickly when the move exceeds the breakeven, and index moves of 2% to 3% on Fed or inflation days do happen.
Volatility does not crush. The example assumed both expiries returned to 13%. If the decision changes the outlook and implied volatility rises across the curve, the back leg helps, but the front’s realized move usually dominates.
Unscheduled events. Event-variance models only price the calendar you put in. Tariff announcements, geopolitical shocks, or a bank failure land on “base” days and invert the whole curve at once, as the April 2025 and March 2020 rows show.
Gap risk and short-dated gamma. With daily expiries, a large share of index option risk sits in contracts with hours left. Their gamma near the strike is extreme, so small errors in the event estimate become large P&L.
Calendar-day versus trading-day conventions. Mixing the two misprices weekends and holidays and can make an expiry look rich or cheap when it is not.

The short version

Implied volatility forms a curve across expiries. Usually it slopes upward; in stress it inverts, which in Cboe’s data happened on about 8% of days since 2009, most sharply in August 2015, February 2018, March 2020, and April 2025. Because variance adds over time, two expiries give forward volatility, and an expiry with an event gives event-day volatility; in our FOMC example, 13% and 19% options imply a decision day priced at 23.5% volatility, or an average move of about 1.2%. Since 2022, daily SPX expirations let traders isolate those days, and Cboe data show short-dated S&P 500 volatility rising into Fed decisions and falling after. Calendar spreads trade that shape, but one spanning an event is mainly short the event move, and it loses when the move beats the price.