VIX backwardation, contango and the volatility futures term structure

The VIX curve articulates the spot index and nine CFE VIX futures expirations. Its shape, contango or backwardation, provides a regime signal distinct from the spot level and historically more robust for identifying acute stress phases.

Contango dominates 79% of trading days from 2013 to 2026; backwardation, rarer, has accompanied every S&P 500 decline of more than 10% over the period, without ever preceding it by more than a few days. What follows reconstructs the curve mechanics, its historical configurations, and the structural impact on volatility ETPs.

The VIX curve: a forward instrument, not a tradable underlying

Spot VIX is not directly tradable. Market participants seeking to buy or sell equity volatility use VIX futures, launched in March 2004 by the CBOE Futures Exchange (CFE). Each contract is on the expected VIX level at a future expiration; the market continuously quotes nine consecutive monthly expirations (VX1 to VX9), from the current month to a twelve-month horizon. A related dataset: The implied-volatility record.

The curve formed by these nine expirations is not a simple expectation. It reflects the equilibrium between buyers (institutions hedging against a future volatility spike) and sellers (systematic players harvesting the structural risk premium). The price of a VIX future at expiration T₁ thus embeds both the expected VIX value at T₁ and the risk premium sellers demand for carrying spike risk.

Three conventions are needed to read the curve correctly. First: VIX futures settle in cash on the value of the Special Opening Quotation (SOQ) of the VIX on the morning of the Wednesday thirty days before the third Friday of the following month (usually the third Wednesday of the expiration month), not on the prior day’s close. This particularity generates sometimes substantial fixing effects. Second: contracts expire sequentially, mechanically rolling the curve leftward at each monthly expiration. Third: liquidity decreases with maturity; VX1 and VX2 concentrate 81% of volume (August 2024 to April 2026), and beyond VX5 bid-ask spreads can widen materially.

Contango and backwardation: definitions and frequency

When the curve is upward-sloping (spot VIX < VX1 < VX2 < … < VX9), the regime is called contango. This is the dominant regime: volatility sellers demand a rising premium to carry risk over longer horizons, which lifts the curve. Contango signals a market anticipating moderate future volatility and compensating option sellers for their commitment.

When the curve is downward-sloping, spot VIX > VX1 > VX2 > …, the regime is called backwardation. This regime is rare and associated with acute stress phases: spot VIX leaps beyond the levels the market anticipates on longer horizons, signalling that participants assess the shock as transitory. Backwardation can be total (entire curve inverted) or partial (only spot VIX > VX1, with the rest of the curve remaining in contango).

Over the 2013-2026 period (contract histories published by Cboe, retrieved 22 September 2026), contango dominates 79% of trading days per Eco3min calculations. The remaining 21% split between partial backwardation (spot VIX above VX1 only, 10%) and a curve fully inverted out to VX6 (3%). Total backwardation rarely lasts more than a few weeks, except in phases of massive repricing: 88 sessions from 24 February to 26 June 2020 (COVID), 47 sessions from 3 March to 7 May 2025 (tariff shock), 46 sessions after Volmageddon in 2018; the August 2024 yen carry trade unwind inverted the front of the curve for only nine sessions, 1 to 13 August, and the whole curve for a single one, 5 August. Related Q&A: our breakdown of the carry trade.

Reading the slope: what curve shape says

Several slope metrics are tracked by volatility desks. The simplest: the VX1 / spot VIX ratio. In normal contango, this ratio sits between 1.04 and 1.08 — VX1 priced 4 to 8% above spot. In backwardation, the ratio drops below 1.00; the extreme episodes saw this ratio fall to 0.75 on 26 March 2020, 0.71 on 7 April 2025 and 0.83 on 5 August 2024, meaning spot 20 to 40% above VX1. The detail is worked through in the assumptions that mislead investors on commodities and gold. A parallel read: our analysis of investing in commodities.

A second metric, more informative: the VX2 / VX1 ratio. When this slope flattens (moving from 1.03-1.05 to 1.01-1.02) without curve inversion, it is an early tension signal. This pattern preceded some peaks, notably February 2018 and August 2024, by one to two weeks. The VX2/VX1 slope flattening is one of the leading indicators tracked by volatility funds. This finding is put in perspective by the stress-signal layer of market regimes.

A third dimension: the VX1 vs VX6 slope (current month vs six months). A gap below 5% in contango (instead of the usual 15-25%) signals structural compression that may indicate, depending on context, either a new normal or a latent stress accumulation. This slope averages 10% from August 2024 to April 2026, against 16% over 2013-2024: low, but far from the 5% threshold. The 2023-2026 volatility compression phase documents this joint spot + curve compression.

Backwardation: historical signal of acute stress

Over 2013-2026, twenty-eight episodes of total backwardation (spot VIX above VX1, VX2 and VX3, at least a month apart) have been observed on the VIX curve per Eco3min counts. Twenty-three of them occurred inside an S&P 500 peak-to-trough decline of at least 5%, but only seven were followed by a further 5% fall within thirty days: the flip arrives inside the decline, not before it. The June 2013 taper tantrum inverted only the front of the curve, for a single session, on 20 June.

Three episodes warrant separate commentary. October 2008: the curve inverted after the House rejected TARP on 29 September 2008 and stayed inverted until end-November (pre-2013 contract histories not reconstructed here). Between 29 September and 20 November 2008, the S&P 500 lost 32%. Backwardation returned briefly in February-March 2009 before the equity bottom formed. February 2018: the front of the curve inverted from 1 February, the whole curve on the 5th (Volmageddon day, VX2/VX1 at 0.75 the next day), and spot did not fall back below VX1 until 7 March, after the implosion of several short-vol ETNs; the S&P 500 lost 10.2% from 26 January to 8 February then rebounded. March 2020: total backwardation from 24 February to 26 June, VX1/VIX bottoming at 0.75 on 26 March; the S&P 500 lost 30.6% from 24 February to the 23 March trough.

The backwardation signal is therefore historically robust for identifying acute stress phases, but it is neither a precise timing signal (backwardation can persist several weeks before the bottom) nor a return-to-normal signal (the return to contango does not mechanically mark the end of the drawdown). It is rather a regime indicator: it confirms entry into a repricing phase without specifying when it will end. For the complete mapping of major spike episodes, see the empirical catalog of VIX spikes 1990-2026.

Roll yield and consequences for volatility ETPs

The curve shape has a mechanical consequence on ETFs and ETNs replicating VIX futures. These products — VXX, VIXY, UVXY (long volatility) or SVXY (inverse) — must roll their positions monthly to maintain exposure. Roll yield is the spread between the price of the contract sold (near-term) and the one bought (next term).

In contango, roll yield is negative for long-volatility ETPs: they sell at the low price (VX1) and buy at the high price (VX2), mechanically losing the slope at each renewal. Over 2009-2018, VXX lost on average 60% per year despite a relatively stable spot VIX, entirely through negative roll yield. Symmetrically, the inverse XIV gained 50% per year by capturing the contango slope — until its February 2018 implosion in sudden backwardation.

In backwardation, roll yield inverts: long-volatility ETPs gain at each roll by selling high (elevated VX1) and buying lower (lower VX2). This dynamic explains why long-vol ETPs like VXX can gain 200-400% during crashes (March 2020: +180% in three weeks), beyond the simple VIX increase. The complete series of historical equity stress cases documents this mechanism episode by episode.

State of the 2024-2026 curve regime

Since the August 2024 peak, the VIX curve has returned to contango between each shock, without staying there. The average VX1/VIX ratio from August 2024 to April 2026 stands at 1.032, below the 2013-2024 average of 1.053, because the curve inverted on 141 of 436 sessions, including 21 sessions of total backwardation in April 2025. The VX6/VX1 slope is at 1.10, against 1.16 on average over 2013-2024. The curve compression is consistent with spot VIX compression and supports the two competing theses documented in the general analytical framework around the VIX and its regimes: 0DTE-driven new normal or artificial suppression through structured short-volatility positions.

A monitoring signal tracked by desks: a flip of the VX2/VX1 slope below 1.00 (partial backwardation) without identified shock on spot would be one of the first signs of latent destabilisation. This flip occurred on 118 sessions between August 2024 and April 2026, almost always with a shock on spot (April 2025, March 2026); as of 21 September 2026 the curve is in clear contango, VX1 at 17.9 for a spot of 14.9 (ratio 1.21) and VX6 at 20.4 (Eco3min calculation on Cboe settles).

Last updated — 6 October 2026