Is CVOL Skew a Leading Indicator of Price Trends in Commodities, Bonds, and Currency Markets?

For these instruments, the implied volatility skew may be akin to a classic sentiment indicator. When investors fear mainly extreme upside risk, it may have signaled that the market was overbought and therefore more susceptible to declines than gains. By contrast, when CVOL was exceptionally negative in these markets, the opposite often proved to be true: fear of extreme downside risk over extreme upside risk often indicated that markets for these instruments were oversold and susceptible to a bounce in prices.

There is, however, another, set of markets where the opposite has generally been true over the past 16 years. These markets include 5Y, 10Y and 30Y U.S. Treasuries, soybean oil, and to a lesser extent EURUSD, JPYUSD and CADUSD (Figures 9-15). In these markets, the CVOL skew tended, more often than not, to correctly anticipate the future direction of price trends. When traders saw more potential for extreme upside risk than extreme downside risk, these markets tended to trend higher. When investors perceived more extreme downside risk than extreme upside risk, these markets tended to trend lower. 

But why would bonds, currency pairs (except for AUDUSD) and soybean oil behave in the opposite way (on balance) with respect to the CVOL skew than markets like metals and energy? For government bond futures markets and for major currency pairs, the answer might lie in the fact that central banks are among the biggest participants in these markets. When central banks begin easing or tightening monetary policy, they tend to keep going in the same direction for a long time – as the most recent tightening cycle has depicts. Moreover, interest rate differentials – and the anticipated evolution of interest rate differentials – between currencies are major drivers of exchange rates. Central banks have a penchant for telegraphing their policy intentions to markets and then acting on them over long periods of time.

Why would AUDUSD behave more like gold, silver, copper, or crude oil/crude oil products? Possibly because AUDUSD may be more heavily influenced by commodity prices, with which it often moves in tandem, than it is by interest rate differentials (Figure 16). 

With respect to soybean oil, its behavior is often opposite to that of soybean meal. For example, soybean oil correlates highly with economic growth in China. Soybean meal does not. Soybean oil can also be an additive to fuel blends, especially for diesel, and has sometimes appeared to act as a leading indicator of movements in crude oil prices. Refiners may buy more or less soybean oil as a function of their perceptions for the availability of crude oil in coming months (our article on the soybean oil-crude oil link).

Finally, there are many instruments for which the CVOL skew appears to have had little bearing on future markets returns. This is true for GBPUSD and most of the agricultural complex including corn, wheat and soybeans (Figures 17-20). While CVOL skew does not appear to have had much bearing on the future direction of most crop prices, it is fascinating in a very different way: for all the crop prices, CVOL almost always skews positive. This is to say that out-of-the-money calls almost always cost more than out-of-the-money puts. The reasons for this and its implications, however, are the subject of an upcoming article in our series on CVOL and its uses. 

Even those who never trade options might want to pay close attention to the CVOL skew and its potential implications for future movements in spot futures prices. Some assets’ spot futures prices tend to move in the opposite direction of the CVOL skew whereas others tend to trend in the same direction.

The diffusion index is calculated as follows:

1. We take the daily CVOL Skew history for the past two calendar years (a total of 500 daily observations). Before CVOL data is available we substitute the highly correlated series 0.15 Delta risk reversal, which is also found in QuikStrike.
2. We apply a rank function in a spreadsheet to each daily CVOL skew observation comparing it to previous 499 days + the day itself). For example, if it’s the most negative ever, it gets a zero. If it’s the most positive ever, it gets a 500).
3. Next, we use “if/then” statements in a spreadsheet to map the 0 to 500 scale into a 0 to 100 scale. This gives us a matrix of 0s and 1s with each observation falling into one, only one, of the buckets from 0-1 to 99-100.
4. Finally, we multiply this matrix through by a vector of the reinvested futures returns over the three-month period after the observation and then divide by the total number of observations in each bucket to give us an average return for each of the 100 buckets in the diffusion index.