Volatility: When Dumb Money Wins?
Sep. 29, 2017
Lynn Sebastian Purcell
Summary
Bitcoin and shorting volatility have been two of the best trades of the year.
There is a chorus of opposition claiming that both are “dumb luck.”.
Is it?
Focusing on volatility, I provide four “dumb money” strategies that work.
Few things anger professional money managers more than dumb investments that make oodles of money. This year, 2017, has seen two of these: Bitcoin and volatility. The former began the year at $995 and is presently (at the time of writing) at $4087, which represents a 410% gain in about 10 months. Shorting volatility using the VelocityShares Daily Inverse VIX Short-Term ETN (NASDAQ: XIV) since the begging of 2017 would have resulted in about 100% return in the same period.
 
Both of these trades are “dumb,” in the sense that they involve no sophisticated sense of the underlying phenomenon, and do not respond to any market events. Those gigantic yields are the result of simple buy-and-hold strategies, and they’ve dwarfed some of the best fund returns. So it’s not surprising to find a chorus of professionals claiming that Bitcoin is worthless, a Ponzi scheme, or in a bubble, and money managers who warn that shorting volatility is like picking up pennies in front of a steam roller in this environment.
These yields and their controversy prompt a simple question: is this really all dumb luck?
I’ve covered Bitcoin elsewhere ( and here too). For the present article, I’ll look at volatility strategies. My conclusion: there are at least four “dumb money” strategies that work, and arguably one that passes what might be called “The Barron’s Challenge.”
Dumb Money: The Barron’s Challenge Barron’s recently ran a piece on the problems with going short on volatility despite the return it would have yielded this year. Effectively, they pose the challenge that any volatility strategy must meet if the yield isn’t going to count as dumb luck.
Pravit Chintawongvanich, the derivatives strategist at Macro Risk Advisors, explains a back-tested strategy for shorting volatility as follows:
[W]e simply simulated always shorting the ProShares Ultra VIX Short-Term Futures ETF (UVXY) with a certain % of one’s money and shorting more when the VIX spiked. The point is that yes, it was possible [to achieve returns that would turn $500,000 into $12,000,000 in five years], but only by taking extreme amounts of leverage and suffering multiple high (60%+) drawdowns.
Two parts of the challenge are thus the following: (1) the strategy must have high yielding returns (the math works out to 89% CAGR, but one might be reasonably happy with a “mere” 50% CAGR), and (2) avoid significant drawdowns.
Chintawongvanich continues:
Most importantly, this strategy would not have survived 2008 … [and] the past 5 years have been a uniquely good time for shorting volatility, causing these strategies to have a misleadingly high Sharpe ratio.
So there are two further points here: (3) smart strategies would survive major market down turns, and (4) they would need to have good risk adjusted numbers (e.g. Sharpe & Sortino ratios) over that whole length of time.
Is it possible to come up with a strategy that satisfies these demands relatively easily? I think so.
How Bad Is Dumb Money?In one sense, The Barron’s Challenge is correct. A simple buy-and-hold strategy would not meet all four criteria, but it’s not as bad as one might think. To simplify, I’ll look at just buying and holding the XIV, since it’s more available to retail investors, and avoids making assumptions about borrowing costs when shorting the UVXY. Using the simulated data from Portfolio123, which mathematically extends performance back to 2006, here is what the buy-and-hold strategy for the XIV would look like.
 Surprisingly, it not only would have survived (criterion no. 3), but would have yielded a 21.23% return—beating Warren Buffett’s Berkshire Hathaway (NYSE: BRK.A) over the same period of time. This is important, because it indicates that fears of the XIV going out of existence (losing more than 80% of its value in a single day) are highly unlikely. It did have a whopping 85% drawdown, (failing criterion no. 2), but that would have been over more than a week, so no going out of existence even during the 2008 crash.
Now let’s look at the risk-adjusted performance of the portfolio on an annual basis.
 Interestingly, it does have fairly nice risk-adjusted numbers: a Sharpe of .46 v .31 of the S&P over the same period, and a Sortino of .76 v .38. So it would pass criterion no. 4 (not fantastic numbers I know, but it does beat the benchmark). Finally, one might gripe about criterion no. 1, as 21% is nowhere near the 89% of the article, and not even close to the “more reasonable” 50% criterion.
This “dumb money” strategy, a mere buy-and-hold approach, clobbers the one used in the Barron’s article, and it’s also more obvious (so it’s certainly dumb). It also beats Buffett over the same period—though, admittedly Buffett has much different problems (this would never work for someone looking to invest $100B). Nevertheless, I think one could say that these points suggest that it “works,” even if I wouldn’t ever pursue it. With respect to The Barron’s Challenge, the tally looks as follows.
Criterion
| Outcome
| 1. A 50% CAGR?
| No
| 2. Reasonable Drawdown?
| No
| 3. Survive 2008?
| Yes
| 4. Nice Risk-Adjustment?
| Yes
|
So the Barron’s Challenge is technically right, but with a strategy a little smarter than their own, it comes pretty close. I mean, that’s just buy-and-hold approach. Can we improve on that? I think so.
Progressively Smarter MoneyLet’s think this one through. Given the criteria of The Barron’s Challenge, we’re going to need a strategy that satisfies the following rules.
Rule no. 1: The portfolio stays away from recessions.
Rule no. 2: The portfolio avoids holding in volatility spikes.
Rule no. 3: It’s ok to get burned a little, as long as it doesn’t cause massive draw-downs.
There are some well-known ways to meet these requirements. Here’s how I’m going to do it.
To meet Rule no. 3, I’ll just use weekly data (I’ll come back to this point in the Final Remarks). This means, the portfolio will only adjust to news events once a week.To meet Rule no.1, I’ll make this an oscillating portfolio. When it’s not in the XIV, I’ll just stick it into a bond ETF. For these purposes, I used the IEF.To meet Rule no. 2, we’ll try some experiments.Dumb Money Strategy no. 2
This strategy just uses a 10-50 EMA crossover on the VIX to trade in and out of the XIV. Here are the returns and stats.
It’s better, clearly, as it returns about 29.5% CAGR. Still, it has a 70% draw-down. The Sharpe and Sortino ratios are better, but I’d really like to see at least the Sortino pop up above one (the present numbers certainly aren’t fund-worthy). Our overall tally is about the same as follows.
Criterion
| Outcome
| | 1. A 50% CAGR? | No
| 2. Reasonable Drawdown?
| No
| 3. Survive 2008?
| Yes
| 4. Nice Risk-Adjustment?
| Yes
|
Dumb Money Strategy no. 3
This strategy uses a 2-3 month EMA unemployment crossover to trade in and out of the XIV. The returns and stats are below.
So we’re pretty close to that 50% CAGR, but it’s max drawdown is still close to 70%. The Sortino ratio does go above one, however, so that’s quite positive. Our overall tally now changes a little.
Criterion
| Outcome
| 1. A 50% CAGR?
| Yes-ish
| 2. Reasonable Drawdown?
| No
| 3. Survive 2008?
| Yes
| 4. Nice Risk-Adjustment?
| Yes
|
Dumb-ish Money Strategy no. 4
If you just combine the two above conditions, you get a better CGAR and risk adjustment (clearly meeting criterion no. 1), but it still doesn’t beat the S&P drawdown. To get the drawdown below the benchmark, it takes a bit of semi-unknown knowledge: use credit-spreads. There are a few to choose from, and I just used the Bank of American credit spread for this: a weighted 20-500 day EMA crossover. Below is the chart and stats.
 
You’ll notice that it looks like there’s two sorts of phenomena in the curve, and that’s because this portfolio is an oscillator (between XIV and the iShares 7-10 Year Treasury Bond ETF (NYSEARCA: IEF)), and it responds to fairly specific conditions (we’ve got compound criteria now). I’m not, as a result, particularly worried about that. Here’s our little Barron’s check-list.
Criterion
| Outcome
| 1. A 50% CAGR?
| Yes
| 2. Reasonable Drawdown?
| Yes
| 3. Survive 2008?
| Yes
| 4. Nice Risk-Adjustment?
| Yes
|
Final RemarksI think I’ve made my general point: there are “dumb money” short volatility strategies that would have worked to varying degrees through the recession, and at least one that arguably passes The Barron’s Challenge (it’s arguable, because it’s not clear how dumb it is). It’s worth asking, at this point, how much confidence we ought to have in these results.
Honestly, I don’t like any of those strategies but the last (and I’m hesitant there), since they rely on single measures, and the sample set for down-turns in volatility is relatively small during this period of time (about 12-14 depending on how you’re counting), even though it does include the 2008 crash as serious test-case. The way to avoid this problem is to use multiple predictors, composed and conjoined with each other in various ways, with a more variable strategy for entering the market. The strategy is not obvious (so it’s not “dumb money”), but here is what one possibility looks like.
That does pretty much what’s asked, and when the strategy is back-tested shorting the UVXY it exceeds the 89% CGAR of the actual Barron’s Challenge, even when including borrowing costs. As a note: what’s pictured above (using only the XIV) turns $100,000 into $20M in 12 years.
Another question to ask is whether the weekly data used in these back-tests is really adequate to the needs of someone seeking to invest their hard-earned money. I’ve looked into it, and it turns out that when you use daily data, you can get better results, even using the same strategies. BUT they’re not so much different. The reason is that even if one limits oneself to buying and selling on market opening, there is still quite a bit of “noise” in these turbulent periods. This is to say, one will end up selling one day at a loss, only to buy the next, and then sell yet again at a loss. If one had just held for the whole week, the losses would have been about the same. It also adds to one’s transaction fees. The real gains tend to occur around downturns which are spaced a couple of weeks apart (like some of those in 2014 and 2015), but again, they are less than one might expect.
Finally, all of these strategies have been strictly algorithmic. This is to say, I have eliminated all human “intuition” from interfering with the trading. Can human reasoning improve these returns, or perhaps prevent some of the greater down-turns?
I’m suspicious about that. One can use human intuition with volatility, but the best way to do that, I suspect, is to have a purely human intuition based strategy—for example, deciding that after any major spike, and testing for macro-economic conditions, one will short volatility for a certain period of time, say three weeks. The reason I’m suspicious about human intuition adding much to the algorithms is that the algorithms tend to get their best returns when human intuition would tell us that the coast isn’t clear yet, or that a disaster is looming just three days away (e.g. the VIX can’t possibly fall below 10!). |
