Goldman's Take On Enhancing Returns In A Yield Deprived World

Traders and asset managers across the world have found themselves in a market deprived of yield. We have our central bankers and their policies of zero or negative interest rates to thank.

While the Fed is on track to raising rates for the first time since embarking on its 7-year journey of zero interest rate policy, it is confronting market that has gotten so accustomed to an environment of yield repression.

Asset pricing models have long adapted to this new financial zeitgeist; it is also safe to assume that most financial assets have never been more sensitive to even the tiniest shifts in interest rates. Huge dangers are associated with this dynamic, and central banks and financial market regulators are mostly behind the curve in preventing what is known as a "fat tail" event.

Having published our market outlook for August and beyond, we thought it would be appropriate to introduce a lesser talked about subject to the picture.

This piece may be a little technical for some readers, but we feel a full read through is worthwhile.

The Gold Rush For Yield

The quest to find every marginal basis point of return has led the smartest minds to venture where few dare to. There has perhaps never been a time with this abundance of money being left clueless on where next to pour into for that extra basis point of alpha. This scares us.

 There has been a net outflow from equity funds since the start of 2015. Institutional demand has been extremely weak, and chances are that such funds have rotated into the high yield bond market.  Chart courtesy of BofAML

There has been a net outflow from equity funds since the start of 2015. Institutional demand has been extremely weak, and chances are that such funds have rotated into the high yield bond market.

Chart courtesy of BofAML

Evidence of this phenomenon is widespread. High frequency trading (HFT) is one way which traders aim to game the market which they operate in, investing extraordinary effort and capital in cutting edge technology which puts them microseconds ahead of the broader market, earning them millions of dollars across the thousands of huge trades they execute.

Another way the less adept of us try to chase diminishing returns is through high yield securities, also known as junk bonds. The HY market has never been more saturated with both institutional and retail capital. Fund inflows into these inherently risky tranche of corporate debt have been piling up, even while funds are covering exposure in equities.

The Silent Supremacy Of Selling Volatility

There are various other ways in which zero-yielding cash tries to gain alpha. One such way, and one which we feel has been one of the best and most consistent strategies to enhance returns, is to sell short-term volatility (vol) on U.S. equities.

This often misunderstood and underrated strategy has generated an impressive overall return through the last decade. The consistency of this strategy is what attracts us, along with its relative simplicity both in theory and practice.

Over the weekends, none other than the venerable "vampire squid" investment bank Goldman Sachs, published a piece highlighting how this strategy of short-term vol selling has been so effective in times when the market is starved of yield.

Goldman's point is clear. By selling front-month vol (through weekly VIX futures contracts), one is able to ride on the back of exponential decay to spot VIX, earning the highest possible carry while rolling the position each week. Over time, future vol will most likely be expensive to spot vol, and spot vol should overall remain suppressed in current market conditions.

For a long only equity portfolio, hedging can be done by buying VIX futures to protect against downside in equities; vol rises when equity prices fall. Flipping this over, it is possible to earn a low but consistent return by selling VIX futures. According to Goldman, this strategy works approximately 70% of the time.

In April, we published a piece detailing the effects the end of the Fed's QE affected the volatility markets in a profound manner. While vol often remains low for much of the time, underlying convexities have caused the VIX to become more unpredictable and difficult to trade directionally.

However, this long-term strategy when implemented correctly, should be able to negate much of the additional time-specific risks associated with shorting the VIX. Again, the key is in the duration of which the strategy is implemented.

 Total returns from selling front-month VIX futures (November 2005 to July 2015) stand at 114.33 points when contracts are rolled monthly. There is a 70% average success rate for analyzed period. Even during high vol periods of 2008, this strategy still managed a success rate of 42%.  Chart courtesy of Goldman Sachs

Total returns from selling front-month VIX futures (November 2005 to July 2015) stand at 114.33 points when contracts are rolled monthly. There is a 70% average success rate for analyzed period. Even during high vol periods of 2008, this strategy still managed a success rate of 42%.

Chart courtesy of Goldman Sachs

Over the past 10 years (including the high vol events of 2008 and 2009), this strategy of selling short-dates vol has yielded more than 114 index points. We are impressed. And now that weekly VIX futures are available, smoother returns are possible over a shorter length of time.

We feel this strategy can be extremely cost effective for funds with large holdings of HY and IG debt. Correlations between corporate fixed income returns and those of equities is substantial to make this strategy viable. Other notable advocates of vol selling is Bill Gross when he was still head of PIMCO.

Low Vs. Cheap, High Vs. Expensive

One last point we wish to add is that there is a marked difference between low vol and cheap vol, high vol and expensive vol. In so many cases, traders buy vol when it is low, only to have them decay away. Traders also sell vol when it is high, only to see it edging higher.

Vol is often expensive when low - reflecting the fundamental element of the premium one has to pay for excess returns over a risk free rate. This premium should be sold, because volatility is expensive most of the time when equity returns are positive.

Conversely, vol is often cheap when rising as equity returns are negative and downward momentum is strong. This is when vol should be bought; this is especially so because spot vol is often the most sensitive to the initial move while further-dated futures are less sensitive.

Goldman On Selling Volatility As The Best Yield Enhancing Strategy

The Short Story: Short VIX futures index +56% YTD

If the longs use VIX products as hedging instruments, then why would anyone take the other side? Because, being short volatility can be very profitable. The S&P 500 VIX Short-term Futures Daily Inverse Index (SPVXSPI) tracks the profitability of being short a constant maturity 1m VIX future and is the benchmark for ETPs such as the XIV and SVXY. Year-to-date this short vol index is up 56%.

 Since the start of 2015, the index which tracks the inverse of short-term VIX futures has rallied 55.9%, handily making it the top performing benchmark, if one can call it so, across all other comparable assets. Several inverse VIX ETFs use this index as a benchmark. Despite equity returns being mostly muted YTD, selling short-dates vol has proven to be immensely profitable.  Chart courtesy of Goldman Sachs

Since the start of 2015, the index which tracks the inverse of short-term VIX futures has rallied 55.9%, handily making it the top performing benchmark, if one can call it so, across all other comparable assets. Several inverse VIX ETFs use this index as a benchmark. Despite equity returns being mostly muted YTD, selling short-dates vol has proven to be immensely profitable.

Chart courtesy of Goldman Sachs

In low vol environments VIX futures tend to trade above VIX spot and futures typically roll down the curve to settle at VIX spot.

Short VIX futures strategies profit from the contango in the VIX futures curve. The steeper the VIX term structure, the higher the (futures-spot VIX) basis, and short VIX strategies tend to be profitable as futures roll down the curve. There are many investors who try to profit from this well publicized phenomenon: sell a VIX future, capture roll down, do it again (wash, rinse, repeat).

Prior to VIX Weeklys if you wanted to capture the roll-down you might have sold the front-month contract and hoped for the best. Short vol investors know that putting all of your eggs in one basket can be a risky strategy.

VIX Weeklys may provide more flexibility with investors positioning for the roll-down a bit week each week by simply spreading out their monthly trades. Instead of selling $100 on the front month VIX future an investor might sell 1/4th of the notional per week which may help smooth the return profile. The VIX often mean reverts quickly so if one contract expires in the red, the other contracts may pick up the speedy mean reversion and end in the green.

On the tactical side, we could see more investors positioning for a swift decline in volatility post an event (FOMC for example).

Shorter-dated VIX futures track VIX spot more closely

A one-month VIX future has a beta of 0.44 to the VIX. For example, if the VIX moves up a vol point, a future with one-month left to expiration tends to move up a little less than half as much, or 0.44 vol points.

 The advent of weekly VIX futures means that traders are now able to slide down the contango curve faster than ever before. As expected, the closer to spot a futures contract is, the higher its beta to spot. In the case of the VIX, a 1-week futures contract has 1.4x the beta of a 1-month contract, allow that much more decay in a quarter of the time, assuming the trader levers up 4 times the notional.  Chart courtesy of Goldman Sachs

The advent of weekly VIX futures means that traders are now able to slide down the contango curve faster than ever before. As expected, the closer to spot a futures contract is, the higher its beta to spot. In the case of the VIX, a 1-week futures contract has 1.4x the beta of a 1-month contract, allow that much more decay in a quarter of the time, assuming the trader levers up 4 times the notional.

Chart courtesy of Goldman Sachs

Higher betas for shorter-dated tenors. The beta between a future with one week to expiry and the VIX has been 0.64 or 1.4x higher than a one-month future and 2.2x more sensitive to VIX moves than a future with three-months remaining to expiration.

Many exchange traded VIX products are benchmarked to constant maturity VIX futures. As a cross check on our reaction function we create constant maturity one- to six-month VIX futures each trading day and estimate the betas back to VIX changes.  The results are very similar, with a constant maturity one-month VIX future having a beta of 0.45 to VIX changes and threemonth futures at 0.27, very close to what our reaction function would have predicted (1m: 0.44; 3m: 0.29).

Shorter-dated VIX futures track S&P 500 more closely

The beta between daily changes in the VIX and daily S&P 500 returns has been -1.2 using data back to 2004. A beta of -1.2 implies that for every -1% decline in the S&P 500 we would expect the VIX to go up by 1.2 vol points (say from 15 to 16.2).

 Another advantage of rolling weekly VIX futures over monthly contracts is the substantially higher beta (1.4x) the former has to the S&P 500 equity index. The historical VIX-SPX beta going back to 2004 stands at -1.2, implying that VIX should rise 1.2 vol points per 1% decline in the SPX. A tighter linkage to the underlying means that traders have more control over the the strategy. It is interesting to note that longer-dated VIX contracts can be hardly affected by more mundane moves in stocks.  Chart courtesy of Goldman Sachs

Another advantage of rolling weekly VIX futures over monthly contracts is the substantially higher beta (1.4x) the former has to the S&P 500 equity index. The historical VIX-SPX beta going back to 2004 stands at -1.2, implying that VIX should rise 1.2 vol points per 1% decline in the SPX. A tighter linkage to the underlying means that traders have more control over the the strategy. It is interesting to note that longer-dated VIX contracts can be hardly affected by more mundane moves in stocks.

Chart courtesy of Goldman Sachs

The beta of a one-month VIX future to S&P 500 returns is -0.60, roughly one-half the sensitivity between the VIX and market returns; about 1.5x that of a three-month future (-0.41) and 2.1x a six-month future (-0.28).  We make two important points here: (1) you cannot trade spot VIX and the betas between the tradable VIX futures and the market have historically been much lower; (2) the betas fall off dramatically as you move further out in the term structure.

As a VIX future approaches settlement, its sensitivity to S&P 500 returns grows exponentially. The beta of a VIX future to S&P 500 returns moves from -0.6 on a one-month future to -0.74 on a two-week and -0.86 on a future with one-week left before VIX settlement and -1.15 with one-day left to maturity which approaches the beta of VIX spot (-1.2).

Understanding The VIX Term Structure

For us to fully appreciate this strategy of selling short-term vol for income, we need to understand how the term structure of the VIX affects the rate of roll-down (decay), so we can better strategize.

What Goldman explains below is that most of the roll-down to spot in the front-month VIX future occurs in the last week before expiry. The roll-down can be likened to the theta (time decay) of an American option, where decay happens at an exponential rate as the option speeds to expiry.

Goldman On The Killer Contango: The Low-Down On The Roll-Down

VIX Weeklys: Understanding the front-end of the VIX futures curve

We expect most of the VIX Weeklys flow to be traded one-month and in. Therefore, understanding how VIX futures move at the short-end of the curve will be critical for successful trading.

We perform two simple studies to put historical moves on short-dated VIX futures in context:

  1. Quantify the monthly gain on a short one-month VIX futures position. Said another way: How much does a 1m VIX future tend to roll down the curve due to contango?
  2. Estimate the typical trading pattern of a VIX future over the final one- and two-weeks of trading. What proportion of the roll down happens over the final week?

Study Methodology: Enter a short 1m VIX futures position (nearby contract). The trade is held for one month and rolls T-1 business days prior to the next official VIX settlement date (Tuesday before Wednesday VIX settlement).

 This chart shows how much a 1-month VIX futures contract rolls down the curve in a typical contango. Much of the slide happens over the last week of trading. 2015 has seen an extraordinarily steep contango, which has meant that the average decline in the front-month contracts stands at 2.3 vol points per month YTD, 230% the long run average measured since July 2005.  Chart courtesy of Goldman Sachs

This chart shows how much a 1-month VIX futures contract rolls down the curve in a typical contango. Much of the slide happens over the last week of trading. 2015 has seen an extraordinarily steep contango, which has meant that the average decline in the front-month contracts stands at 2.3 vol points per month YTD, 230% the long run average measured since July 2005.

Chart courtesy of Goldman Sachs

For example, a short July VIX futures position would have been entered on June 16, 2015 (the last trading day before June settlement) and exited on July 21, 2015 (the last trading day of the July contract). Our numbers should tell us the profitability of passively selling a one-month VIX future. Our first position is entered in October 2005, when successive monthly futures contracts became available. We measure our P/L in vol points, for example, if the VIX future starts at 20 and drops to15 the short would make $5 per future traded (5 vol points).

A few highlights:

The average monthly decline on the front-month VIX future has been 0.98 vol points back to 2005. Over the 117 months from November 2005-July 2015, selling a front-month VIX future was successful 70% of the time with an average monthly gain of $0.98 (median: $1.47) for a cumulative gain of $114.3. Profitability varied widely by year.

2015: The average monthly roll-down so far in 2015 has been 2.3 vol points or about 2.3x the long run average. Steep contango has led to strong performance on short VIX futures strategies so far in 2015. As mentioned above, the S&P 500 VIX Short-term Futures Daily Inverse Index (SPVXSPI), the benchmark index for ETPs such as the XIV and SVXY, is up 56% YTD. The last time rolls have been this high was in 2012 when the roll-down averaged 3 vol pts per month and the annual performance of the SPVXSPI was +162%.

70% of the profitability on a short VIX futures position has come over the last week of trading. A more granular analysis shows that the average decline in a one-month VIX future in the week leading into VIX settlement has been 0.68 vol points or 70% of the total one-month decline back to 2005.

The average decline in the front month VIX future from ten days prior to VIX settlement to the last day of trading has been 0.46 vol points over our entire sample window.

Caveat: While 70% of the VIX roll-down occurred over the last few days leading into settlement that number has varied widely by year and moving from five business days out to six or seven will change the numbers.

2015: Large roll-down over the last week of trading. Of the average 2.3 vol point VIX futures monthly roll-down in 2015, 1 vol point (44%) of that has come over the last five days prior to settlement. Our point here is that the roll is heuristically similar to the theta of an option. A lot of the profitability on a short option trade can happen as the option speeds into expiration. In our example, the future converges lower to VIX spot.