Skip to content

VIX Futures Term Structure and the Expectations Hypothesis: summary and key takeaways

The paper checks whether the steepness of the VIX futures curve forecasts later moves along the curve. For the short end the slope gets the sign right but exaggerates the move. For the long end it fails on sign and size. Spread trades built on this pattern made excess returns, which the paper links to ETN-driven demand for long VIX exposure.

Asensio, I.O. · 2020 · Quantitative FinanceSummary updated 20 November 2026
Published2020Journal article
Asset focusVolatility/VIXVIX term structure
TypeResearchVolatility & VIX
Reading time5 minthis summary

VIX Futures Term Structure and the Expectations Hypothesis at a glance

The question
Does the slope of the VIX futures term structure correctly predict the later evolution of short-dated and long-dated futures prices, and what keeps any deviation in place?
Dataset & period
Daily closes of the SPX and VIX indices, VIX futures and VIX option implied volatility, January 2006 to December 2018, sourced from CBOE through Bloomberg. Futures tenors cover the first 8 months of maturities. The driver regressions use non-overlapping weekly data, with a sub-period of 2011-2018 for the ETN era.
Method
The paper adapts the Campa and Chang framework to the VIX curve. Test 1 regresses the average subsequent change in the near futures price, in squared terms, on the long-short spread. Test 2 regresses the one-month change in a long-dated contract on the same spread. The hypothesis holds if the intercept is zero and the slope is one. The author then forms exposure-neutral and beta-neutral spread trades and regresses their profits on equity factors, VIX futures open interest, hedge fund capital, carry trade returns and last week's trade return.

Key findings of VIX Futures Term Structure and the Expectations Hypothesis

  • In test 1 every slope coefficient is positive, so the spread forecasts the sign of the later short-end move but not its size. For the 8-month curve the slope is 0.619 with an r-squared of 0.25.
  • The slope coefficients move further from one, and the fit worsens, as the tenor gets shorter. The paper links this to mean reversion in the VIX.
  • Test 2 rejects the hypothesis at every tenor. Slopes are negative throughout, for example -0.219 in the 8-1 on 8 test, so the spread forecasts neither sign nor size of the long-end move.
  • The 8-month curve slopes upward 86% of the time (2006-2018). The paper sees the negative returns on long futures positions as the trading counterpart of these test results, with bigger losses at short tenors.
  • Beta to the VIX falls with tenor: 0.60 for the first future and 0.16 for the 8-month future. Short-dated contracts are therefore favored for building VIX ETNs.
  • Above transaction costs, 14 of 16 exposure-neutral spread combinations are profitable, against 9 of 16 beta-neutral combinations. Equity factors do not explain the beta-neutral profits.

Robustness, as reported in the paper

  • Returns are shown on both arithmetic and geometric bases because daily compounding can bias performance figures either way.
  • Spread trade returns are reported before an estimated transaction cost of 0.01% per day. The paper checks which combinations stay profitable after that cost.
  • The driver regressions are repeated for the ETN-era sub-period. Open interest is significant in 4 of 8 combinations for the full sample and in 7 of 8 combinations in that sub-period, and the intercept is rarely significant.
  • Weekly variables in the regressions are tested for stationarity and checked for multicollinearity through correlations.

Caveats and limits

  • The test setup assumes stochastic volatility, zero correlation with the underlying and no volatility risk premium. The paper does not test the last assumption directly.
  • Explanatory power of the driver regressions is low, with an average adjusted r-squared of 0.05 for the beta-neutral trades, and the intercept is often significant, so omitted factors matter.
  • Headline spread returns are shown before transaction costs, and the sample covers a limited number of stress episodes (inferred).
  • The ETN-demand explanation is supported by regression sign and open interest patterns, not by a direct test of causality (inferred).

Items marked (inferred) are our reading of the design, not statements by the authors.

Why it matters

The paper shows that a steep VIX futures curve has not been an unbiased guide to where short-dated futures end up. It also points to structural ETN demand, rather than equity risk compensation, as a reason the short-end basis stayed elevated. This is useful context for anyone studying roll yield, curve trades or vol-selling products.

Questions about this paper

Does the VIX futures curve predict where short-dated futures will go?

Only the sign. The paper finds the curve exaggerates the size of the later short-end move.

Why do the spread trades earn excess returns according to the paper?

The paper attributes the profits to an elevated short-end basis driven by long VIX demand entering through ETNs. It finds that equity market factors do not explain the beta-neutral profits.

What does the paper say about the long end of the curve?

Every tenor rejects the hypothesis. The spread forecasts neither the sign nor the size of the next move at the long end.

Where can I read the original paper?

The original is at its publisher: https://doi.org/10.1080/14697688.2019.1684549. We link to the paper and do not host the PDF.

Follow new quant research

Download MacroFinance market data as CSV or Excel, or pull it by API, and test the ideas in these papers yourself. New summaries are added as we review them.

Create a free account
Summaries are written by MacroFinance from the published papers, in our own words, for information only. They are not investment advice and may omit detail. Papers are not hosted here: each page links to the original at its publisher.