On September 9, 2026, the tape looked stressed. The S&P 500 closed down 0.48%, the Russell 2000 shed 1.30%, oil pushed toward $100 a barrel on renewed Iran tensions, and a fresh round of U.S.–Canada tariff proclamations hit the wires. The VIX rose 4.58% to 16.46. If you were deciding whether to carry SPY exposure through the night, that stack of headlines argues for caution.
The volatility curve disagreed. And the curve — not the level — is the part worth reading before you hold something you cannot trade for seventeen hours.
The number that matters is a ratio, not a level
The VIX measures expected S&P 500 volatility over the next 30 days. VIX3M measures the same thing over roughly 93 days. Divide the first by the second and you get the implied volatility term structure ratio — IVTS for short.
Below 1.00, the market prices more uncertainty three months out than one month out. That is contango, and it is the normal state of the world. Above 1.00, the near term is priced as scarier than the far term. That is backwardation, and it means something has gone wrong right now rather than someday. Readings between roughly 0.998 and 1.002 are borderline and should be treated as noise.
A VIX of 16 tells you the absolute price of insurance. The VIX term structure tells you the shape of demand for it — whether the market is worried in general or worried tonight. For an overnight holder, that distinction is most of the decision.
Four zones of the curve and what each implies overnight
| IVTS (VIX ÷ VIX3M) | Curve shape | What it usually means | Overnight read |
|---|---|---|---|
| Below 0.85 | Steep contango | Near-term calm priced at a discount to the horizon | Gaps skew small; the real risk is oversizing into quiet |
| 0.85 – 0.95 | Normal contango | The base case — the two-year median sits near 0.87 | Ordinary two-sided risk; standard sizing |
| 0.95 – 1.00 | Flattening | Short-dated protection is being bid; stress is building | Worth trimming ahead of a catalyst, not after |
| Above 1.00 | Backwardation | Acute near-term stress | Highest-variance environment in the sample |
What backwardation has actually done since 2010
Since 2010, the VIX/VIX3M ratio has closed above 1.00 on 321 trading days — 7.7% of the sample — spread across 103 separate episodes. Average duration is 3.1 days; the median is 2. Most inversions are brief, shallow, and forgettable.
The tail is where the damage lives.
| Episode | Length | Peak IVTS | SPY, start to end |
|---|---|---|---|
| Feb 24 – Apr 23, 2020 | 43 days | 1.3437 | −13.44% |
| Aug 20 – Sep 1, 2015 | 9 days | 1.3083 | −5.98% |
| Feb 2 – Feb 13, 2018 | 8 days | 1.3267 | −3.43% |
| Apr 2 – Apr 24, 2025 | 16 days | 1.2739 | −3.16% |
| Aug 2 – Aug 7, 2024 | 4 days | 1.1442 | −2.67% |
| Apr 7, 2026 (most recent) | 1 day | 1.0082 | 0.00% |
Across all 103 episodes, the average SPY return from an episode’s first day to its last is −0.67%. That number is small precisely because most episodes are two days long and resolve without incident. The distribution behind it is not small — a handful of episodes account for nearly all of the drawdown.
That is the honest read. Backwardation is not a sell signal. It is a variance signal: it says the range of tomorrow morning’s outcomes just widened, not which direction it widened toward. The same asymmetry shows up in gap data, where headline averages hide identically fat tails — we went through which SPY gap statistics actually hold up in a separate study.
Where the curve sits tonight
At the September 9 close, VIX was 16.46 and VIX3M was 18.87, putting the ratio at 0.8723 — ordinary contango, and day 107 of the current regime. The last inversion lasted a single session, on April 7, 2026.
Two details matter more than the headline number. First, 0.8723 sits almost exactly on the two-year median of 0.8737 — this is a textbook-average curve, not a complacent one. Second, the ratio rose from 0.8548 the prior session. The curve is flattening slightly. Oil, tariffs and long-end yields moved the level of implied volatility yesterday; they have not yet changed its shape.
For scale: the 95th percentile of the past two years is 1.0230 and the 99th is 1.1237. April 2025 peaked at 1.2739. Nothing in the current reading is close to that.
Three ways overnight traders misread the term structure
- Treating contango as safety. Roughly 92% of trading days sit in contango — which means the overwhelming majority of bad overnight gaps also happened in contango, by sheer base rate. The curve tells you the market is not pricing acute stress. It does not tell you the market is right.
- Reacting to a single-day cross above 1.00. The median episode is two days, and the most recent one touched 1.0082 for exactly one session before resolving. Sustained inversion — five sessions or more — is the version that has historically preceded the deeper drawdowns.
- Reading the curve in isolation. Term structure is one instrument on the panel. We’ve covered six indicators that describe overnight risk better than the VIX alone; the ratio belongs alongside breadth and positioning, not instead of them.
Putting it to work before the close
The term structure is weather, not forecast. Its practical use sits upstream of entry: it belongs in the sizing decision, not the direction decision. A flattening or inverted curve is a reason to carry less through the close, to widen the adverse range you’re willing to tolerate, or to skip the night entirely — because the overnight position sizing math changes materially when the expected range doubles.
The curve does not tell you what happens tomorrow. It tells you how wrong you can afford to be tonight.
SpyTomorrow’s nightly pressure readings are built to sit alongside inputs like this one — decision support for a position you have already sized, never a signal to follow on its own. Neither the reading nor the ratio predicts the open. Both help you decide how much of it you want to own when it arrives.
Term-structure figures in this article come from daily Cboe VIX and VIX3M index closes as compiled by this VIX term structure tracker, current through the September 9, 2026 close.
This article is for educational and informational purposes only. It does not constitute financial advice, investment recommendations, or a solicitation to buy or sell any securities. Past performance and historical data do not guarantee future results. All trading involves risk, including the potential loss of principal. Always do your own research and consult with a qualified financial advisor before making trading decisions.
