When the ETNs were introduced in Jan. 2009, we thought they were interesting because we intuitively knew that implied volatility went up when equity prices fell – true negative correlation. A look at the VIX index confirmed as much, the correlation between the S&P 500 and the VIX was -0.37 since inception in 1990 through the ETN launch in 2009, using annual data.
We quickly realized that the ETNs were not tracking the VIX though, and admitted to ourselves that since we didn’t fully understand the difference between the short-term structure and the medium-term structure, we wouldn’t invest.
We were lucky because that decision was near the market bottom and stocks have rallied hard while the VIX dropped. Over the past two years ending on Dec. 31, the S&P gained 45.5 percent while the VIX lost 55.6 percent. The medium-term VIX index fell 33.7 percent and the short-term VIX index fell by an astounding 90.2 percent.
The massive divergence between the performance of the short and medium-term VIX indexes are the result of the steep contango curve currently embedded into the roll yield. The new UBS product seeks to profit from that contango by going 100 percent long the medium-term index and going 50 percent short the short-term index.
The long-short term structure index has data available on Bloomberg dating back five years and the results are quite impressive. Granted, this is a very short time horizon, but it does include the second worst market decline and one of the strongest bull markets in market history.
For the five years ending on Dec. 31, the Long-Short VIX index gained 194.53 percent, or 14.78 percent per year. Annualizing the monthly data shows a standard deviation of 14.78 percent (less than stocks, which were 17.82 percent), for a 1.46 Sharpe Ratio.
In the three years when stocks posted positive double digit gains (2006, 2009 and 2010), the Long-Short VIX also earned double digit returns: 10.89, 24.16 and 55.44 percent respectively. The down year, 2008, stocks lost 37 percent and the Long-Short VIX also posted a double digit return of 16.40 percent.
Interestingly, in the worst months of the crisis, the Long-Short VIX actually lost money. In Sept. and Oct. 2008, the Long-Short VIX index actually lost a cumulative 7.21 percent.
Essentially, the loss related to being short the short-term index swamped the positive results of the mid-term index. During those two months, the short-term index gained nearly 300 percent (which is a loss for short holders) and the medium-term index 63.12 percent – strong, but not strong enough.
It makes sense when you think about the long-short nature of the index, but it may be cold comfort to lose money on a hedge related product during a steep sell-off. There were two other significant draw-downs that both occurred in 2007. The first one was -11.67 percent in the first three months of the year and the second was -8.25 percent in Aug. and Sep. of 2007.
The correlation between stocks and the Long-Short VIX index is very low, only 0.15 against stocks against 0.30 for bonds, using monthly data. The attractive recent returns, intuitive concept, reasonable volatility and low correlations make this strategy worth investigating.
At this point, we need to do much more research on the merit of this strategy. We need to better understand how backwardation could affect returns in futures years and whether this term structure arbitrage could apply to other futures curves. As far as the product goes, the cost seems very reasonable, but we have been loath to invest in ETNs because of their credit risk.
Still, this is a strategy and product worth serious investigation and consideration.