Trading the VIX Futures Roll and Volatility Premiums with VIX Options

• Date: 01 February 2017
• Published date: 01 February 2017
• DOI: http://doi.org/10.1002/fut.21788
• Author: David P. Simon

Trading the VIX Futures Roll and

Volatility Premiums with VIX Options

David P. Simon*

This study examines VIX option trading strategies based on the systematic tendencies of VIX

futures from January 2007 through March 2014. The strategies involve buying VIX options to

exploit the tendency of VIX futures to rise and fall when the VIX futures curve is in

backwardation and in contango, respectively, as well as the tendency of VIX futures ex ante

volatility premiums to spike and then revert to more typical levels. Subject to caveats about the

often wide VIX option bid–ask spread quotes in the first few years of trading, the results

demonstrate that these limited risk strategies are highly profitable. An important factor driving

the results is that long VIX option strategies greatly benefit from a tendency of VIX option

implied volatilities to rise with increases in the volatilities of underlying VIX futures contracts,

as the latter move toward settlement and their volatilities converge to the typically higher

volatility of the VIX. © 2016 Wiley Periodicals, Inc. Jrl Fut Mark 37:184–208, 2017

1. INTRODUCTION

VIX futures were introduced in 2004 to allow market participants to make direct bets on the

implied volatility of S&P 500 index options and have become heavily traded owing largely to

their hedging properties stemmingfrom the strong tendency of implied volatility to rise when

equity indexesfall.

1

Nevertheless,the historical performance of VIX exchange traded products

that roll over long VIX futures contracts has been extremely poor owing to the negligible

predictive power of the VIX futures curve for subsequent (spot) VIX changes and because the

VIX futures curve has been upward sloping or in contango roughly 3/4 of the time since the

introductionof VIX futures contracts.

2

Because the VIX doesnot tend to rise when VIX futures

are above the VIX, VIX futures tend to roll down the VIX futures curve to a lower VIX at

David P. Simon is a Professor in the Department of Finance, Bentley University, Waltham, Massachusetts.

The author thanks the editor, Robert Webb, an anonymous referee and Jim Campasano for helpful comments

and suggestions. The author also gratefully acknowledges the support of Bentley University through a

summer research grant.

*Correspondence author, Department of Finance, Bentley University, 175 Forest Street, Walth am, MA 02452.

Tel: 781-891-2489, Fax: 781-891 -2896, e-mail: dsimon@bentle y.edu

Received June 2015; Accepted March 2016

1

See Szado (2009) for an analysis of the hedging propertiesof VIX futures and VIX options during the financial panic

beginning in 2008.

2

The VIX futures basis, defined in this study as the spread between the front VIX futures contract with at least ten

business days to settlement and the VIX, is in contango 74% of this study’s sample period from January 2007 through

March 2014. This study uses the term VIX to refer to the spot VIX, as opposed to VIX futures. The iPath S&P 500 VIX

ST Futures ETN (ticker symbol VXX), which rolls over front VIX futures contracts, lost more than 99% of its value

from its inception in January 2009 through mid-August 2014. See Alexander and Korovilas (2011) and Whaley

(2013) for more evidence on the poor performance of exchange traded funds that rollover long VIX futures contracts.

The Journal of Futures Markets, Vol. 37, No. 2, 184–208 (2017)

© 2016 Wiley Periodicals, Inc.

Published online 6 May 2016 in Wiley Online Library (wileyonlinelibrary.com).

DOI: 10.1002/fut.21788

settlement and lose their value.Likewise, when the VIX futures curve is in backwardation and

VIX futures are belowthe VIX, the VIX does not tend to fall and VIX futures tendto roll up the

curve to a higher VIX at settlement. Studies, such as Simon and Campasano (2014),

demonstratethat shorting VIX futures contracts when the curve is sufficiently in contangoand

hedging the risk of a volatilityspike with short S&P futures contracts and buying VIX futures

when the curve is sufficiently in backwardation and hedging with long S&P futures contracts

are highly profitable trading strategies from 2007 through 2012.

The present study examines whether the lack of forecasting power of the VIX futures

curve as well as transitory spikes in ex ante VIX futures volatility premiums relative to actual

S&P 500 index volatilities can be exploited with outright long VIX option positions that enjoy

limited risk. An important factor affecting the profitability of these strategies is the extent to

which VIX options typically are overpriced. The overpricing of S&P 500 index options has

been studied far more extensively than the overpricing of VIX options. Studies including

Bakshi and Kapadia (2003), Broadie, Chernov, and Johannes (2009), Coval and Shumway

(2001), and Eraker (2009) demonstrate that S&P 500 index options build in substantial ex

ante volatility premiums relative to recent actual volatility. For example, Eraker (2009)

reports that from 1990 through 2007 the VIX averaged 19%, whereas the annualized standard

deviation of S&P 500 index returns averaged 15.7%. This large differential caused S&P 500

index option selling strategies to be very attractive on a risk-adjusted basis over this period.

3

Because the VIX reflects model-free implied volatilities of S&P 500 index options and

because the VIX futures curve frequently is in contango, VIX futures contracts typically build

in more substantial ex ante volatility premiums than S&P 500 index options or the VIX.

By contrast, only a few studies have focused on the potential overpricing of VIX options,

which were introduced in April 2006.

4

Song (2012) examines zero-delta long VIX option

straddles and delta-hedged long VIX option positions and finds that they earn significantly

negative returns.

5

Barnea and Hogan (2012) examine the existence of VIX option volatility

premiums by creating synthetic variance swaps on VIX futures contracts. They find that the

realized variance of VIX futures contracts is on average 3.26% below variance swap rates

implicit in VIX options. More recently, Huang and Shaliastovich (2014) examine monthly

delta-hedged VIX option returns and find significantly negative returns from 2006 through

2013that tend to be greater for close to the money VIXoptions, which have more vega exposure,

and when the VVIX—a measure of the implied volatility of 30-day VIX options—is high.

6

The present study assesses whether VIX options tend to be overpriced by examining

unconditional delta-hedged VIX option five-business day returns and then examines the

3

Coval and Shumway (2001) demonstrate that zero beta one-month S&P 500 index straddles lost 3.2% of their

values per week from January 1986 through October 1995. Bakshi and Kapadia (2003) show that delta-hedged long

S&P 500 index call options with 31–60 days to expiration that were at the money to 2.5% out of the money lost on

average 4.44% of their values through expiration from 1988 through 1995. Broadie, Chernov, and Johannes (2009)

show that for one-month at the money S&P 500 index puts, CAPM adjusted returns were a highly statistically

significant 22.5% with Sharpe ratios of .23 from August 1987 through June 2005.

4

Much of the literature on VIX options, such as Grunbickler and Longstaff (1996), Carr and Lee (2007), and Wang

and Daigler (2011) focuses on the relative explanatory powers of different option pricing models rather than on

volatility premiums and the possible overpricing of VIX options. Other recent studies focus on the information

content of VIX options in pricing uncertainty. Chung, Tsai, Wang, and Weng (2011) examine whether the

information reflected in VIX options is useful for forecasting S&P 500 index dynamics, whereas Vokert (2015)

investigates the predictive power of the implied risk-neutral distribution of the VIX. Tsai, Wang, and Tsai (2015)

examine in a high frequency framework whether VIX option quote changes have predictive power for VIX changes.

5

Song (2012) finds that at the money VIX option straddles that initially are delta-neutral lose an average of 6% per

week and 1-month, at the money VIX calls that are delta-hedged daily lose an average of 12% through expiration.

6

Huang and Shaliastovich (2014) find that delta hedged returns of one-month VIX options held to expiration range

from 3.4% to 22.1%% for calls and from 6.2% to 17.7% for puts. The VVIX is constructed by the CBOE and is

an indicator of the implied volatility of VIX options that have 30 days to expiration.

VIX Option Trading Strategies 185