Is the term structure of CBOE Volatility Index (VIX) futures useful for timing the underlying stock index? In the February 2012 version of his paper entitled “The Relationship between VIX Futures Term Structure and S&P500 Returns”, Athanasios Fassas relates the VIX futures term structure to both contemporaneous and future S&P500 Index returns. He measures the VIX futures term structure as the slope of a best-fit line for VIX (spot value) and closing prices for available VIX futures as a function of time to maturity. He rolls futures such that no contract in the calculation is within two weeks of maturity. He tests relationships between change in VIX futures term structure and S&P 500 Index return via regressions run at frequencies of one day, one week, two weeks, one month and two months, with the sample winnowed in each case so that measurements do not overlap. Using daily closing prices of spot VIX and the six nearest VIX futures with at least two weeks to maturity during late March 2004 through July 2010, he finds that:
- On average over the sample period, the VIX futures term structure is upward with annual slope 1.94 (monthly slope 0.16).
- There is a strong positive contemporaneous relationship between change in VIX futures term structure and S&P 500 Index return for all five measurement frequencies. In other words, an increase (decrease) in the VIX term structure tends to accompany a positive (negative) return for the S&P 500 Index. However, the relationship is asymmetrical, with the index loss associated with a slope decrease tending to be larger in magnitude than the index gain associated with a slope increase of the same size.
- Consistent with the view that a steep upward VIX futures term structure means an overbought market, there is a somewhat negative relationship between change in the VIX futures term structure and future weekly, biweekly and bimonthly S&P 500 Index returns.
In summary, evidence indicates that there may be a useful negative relationship between the slope of the VIX futures term structure and future S&P 500 Index return.
Cautions regarding findings include:
- The sample period is short in terms of variety of market conditions and is subject to domination by effects of the 2008-2009 financial crisis.
- The data used in the study are more than six years stale.
- The testing methodology is in-sample, using all data to draw conclusions. An investor operating in real time using only inception-to-date data may draw different conclusions from the same data.
- The approach does not test any trading strategies and does not account for any trading frictions that would accompany such strategies. The findings of predictability may not be practically exploitable.
