Easy Volatility Investing¶
Motivation(s)¶
There are three known facts about stock market volatility:
Stock market volatility, unlike returns, is predictable.
Changes in volatility are negatively correlated with changes in market prices.
Investors will gladly pay a VRP (Volatility Risk Premium) to reduce their exposure to volatility.
The only way to invest in volatility prior to the introduction of ETFs and ETNs was through trading options, futures, or variance swaps. Due to the widespread misbelief that profits from the VIX futures markets come from the roll yield, existing strategies have mostly ignored the VRP.
Proposed Solution(s)¶
The author proposes strategies that exploit the predictability of volatility as an asset class consisting of XIV, VXX, ZIV, and VXZ. Other ETNs such as XVIX can be replicated using the proposed four ETNs. There are ETFs such as SVXY and VIXY that offer similar exposure but with less liquidity and shorter trading history.
The only differences between the four ETNs are the leverage and term length. (XIV, VXX) and (ZIV, VXZ) are paired inverses. The former covers short term (e.g. one to two months) futures positions while the latter invests in medium term (e.g. four to seven months) futures positions. Consequently, (XIV, VXX) have higher roll yield, VRP, and volatility than (ZIV, VXZ). Note that inverse pairs are not necessarily inversely proportional due to volatility drag. For example, VXX has a massive loss of value over some period, but XIV may not experience a corresponding massive gain in value.
The different strategies are
Buy and Hold,
Momentum,
Contango-Backwardation Roll Yield,
Volatility Risk Premium, and
Hedging.
The obvious mean-reversion strategy has been excluded because it depends on external effects (e.g. global financial crisis) rather than the intrinsic properties of XIV, and the author was unable to find a profitable mean-reversion trading rule based on moving average crossovers that came close to matching trend-following rules.
Evaluation(s)¶
Throughout the analysis, the author ignores the possibility of inefficiencies in the term structure pricing because accounting for that requires outguessing the market. Instead, the author reasons that the VRP is an easier target because getting paid to take on some risk is not an inefficiency.
To establish that there is a VRP in the VIX futures market, two nearest VIX futures were weighted so that the mean of their settlement days is 30 days. It turns out this 30 day prediction ordinarily overestimates the actual VIX, which indicates there is a VRP.
The author used historical prices provided by the CBOE and interpolated missing months to create a hypothetical back history of XIV and the other ETNs had they existed from 2004 to 2013. The mean-variance (risk to return) chart of each strategy verifies that a combination of the four ETNs can achieve any level of risk from below-bond risk of about 3% to returns as high as Apple at about 45%. Simply incoporating VRP into an existing portfolio have increased the Sharpe ratio and reduced maximum drawdowns.
There are a few caveats with the proposed strategies. The volatility drag cost 10% to 20% per annum and could occasionally rise to 80% per annum. The amplitude of volatility cycles is increasing with time, which means volatility drag is increasing as well hence reducing the compounded returns. When the term structure cycles between contango and backwardation, constantly switching between XIV and VXX may result in an overall loss. The correlation between VRP and roll yield may disappear when the VIX frequently reverts to its mean. A transition between bull and bear regimes affects all of the strategies, and a sudden drawdown will impact each strategy according to their risk.
Future Direction(s)¶
The roll yield and VRP strategies are quite simple, so would a recurrent neural network be able to come up with this rule?
Question(s)¶
Why don’t more people make use of VRP instead of index funds?
Analysis¶
VRP and Roll Yield strategies can exploit market volatility to achieve high returns as follows:
If the 5 day moving average of (VIX - HV10) > 0, go long XIV else go long VXX.
If the 10 day moving average of VXV / VIX > 1, go long XIV else go long VXX.
However, one should be cautious of going long VXX. See VIX Central for a visualization of the VIX term structure. If more leverage is desired, the equivalent of XIV with options is SVXY, albeit with lower volume.
The author assumes XIV is a random walk since none of the various statistical tests (e.g. variance ratio, Augmented Dickey-Fuller, Phillips-Perron) could reject that claim. This overemphasis of statistical significance contradicts the profitability of contrarian strategies. Another controversial point is the claim that prices grow geometrically, which seems to question why one needs stop-losses. Nevertheless, the analysis is still reasonable since VIX tends to be mean-reverting.
This strategy is no longer feasible since Credit Suisse decided to end its volatility security. However, the VIX term structure is still useful as an indicator.
Notes¶
The Lure and Intrigue of Volatility¶
CBOE’s VIX is a measure of the market’s expected volatility over the next thirty days.
It tends to be mean reverting: extreme values are more likely to be followed by less extreme values.
A simple mean-reversion trading strategy based on moving median crossovers would produce high returns if VIX is tradable.
One crossover strategy is to go long when the VIX crosses below the 11-day simple moving median and go short otherwise.
It is negatively correlated with changes in the SPX.
CBOE’s VXV measures the market’s expected volatility over the next three months.
The ratio of VXV / VIX is more important than VXV itself.
The XIV inversely tracks VIX changes, but it ends up higher than where it started while VIX ends up roughly where it started.
VIX Futures Market¶
VIX futures can be regarded as a form of prediction on where the VIX will close on the day they expire and are settled.
The VIX futures price reflect the market participants’ best collective guess as to what the VIX index will be at settlement because the futures market is a zero-sum game.
The term structure is a chart displaying the current closing price of VIX futures predictions (i.e. contracts).
When the term structure is rising, it is called contango.
When the term structure is falling, it is called backwardation.
The correlation between the VRPO and VRPF is 0.56.
The VIX already has a risk premium built into it, the VRPO.
The VRPF is an additional premium on top of the VRPO.
Roll Yield¶
The roll yield is the difference between the spot VIX and the futures price.
It may pay or cost a small amount every day as the two converge.
The futures price and the spot VIX (i.e. the quoted VIX) must converge to each other on the settlement day to avoid a risk-free arbitrage opportunity.
The roll yield is measurable and includes the VRP.
The VRP can only be predicted and is only known at settlement date.
If there is no VRP, then there is no profit.
The roll yield is positive when the term structure is in contango and negative when in backwardation.
The average roll yield since 2004 has been 5% per month and can rise to 8% during periods of low market volatility.
XIV Dynamics¶
VIX futures are predictions of the VIX and any VIX predictability is already built into the futures prices.
Predicting XIV is not the same as predicting the VIX—it is the same as predicting the predictions of the VIX.
According to the autocorrelation of VIX and XIV for lags 1 to 5, using VIX to predict XIV is not a profitable strategy.
The VIX is negatively autocorrelated for the first three lags, which means that VIX changes can be predicted 3 days hence; but XIV has no significant autocorrelation.
Since XIV shorts two futures contracts with different settlement dates, the expected roll yield is a constant maturity contract with a fixed duration of 30 days.
Trading Strategies¶
Buy and Hold XIV
The annualized return can reach 31.4% per annum.
The drawdowns can frequently reach 50%.
Momentum
Hold the single ETN that has the best return as measured over the last \(k\) days; if all measured \(k\)-day returns are zero, stay out of the market.
The mean-variance chart indicates \(k = 83\) is a good compromise between return and negative correlation.
Contango-Backwardation Roll Yield
Invest in XIV when the VIX term structure is in contango (VXV \(\geq\) VIX) and in VXX when the term structure is in backwardation (VXV \(<\) VIX).
The vratio is defined as the ratio of VXV to VIX.
Volatility Risk Premium
Trade XIV if VRP measure \(> 0\), otherwise trade VXX.
The mean-variance chart indicates HVOL10S is a good compromise between risk and return.
HVOL10 is the current VIX minus the historical volatility calculated over the last 10 days.
To avoid trading whipsaws, apply a 5 day moving average over HVOL10, which results in HVOL10S.
Hedging
BLVDLM is a dynamic hedging strategy aimed at ETN providers.
References
- Coo13
Tony Cooper. Easy volatility investing. National Association of Active Investment Managers, 2013.