Introduction:
Given markets are prone to different regimes, which often last long enough to put proponents of methods antithetical to the environment out of business — or in a psychiatric ward — it is then necessary to have a number of different approaches or frameworks to remain adaptable enough to exploit variant regimes.
What is ‘Short Vol’:
Short vol is a strategy wherein a derivatives trader sells short a put and a call option at typically equidistant strikes out-of-the-money, with the same expiries. This contract works like an insurance contract, hedging the counterparty for moves in the underlying instrument beyond the strikes (plus the premium received), up until the expiry of the options. Once a trade of this strategy is undertaken, the options seller has a short gamma profile, whilst being positively exposed to theta – the rate of decline in the price of an option through time. As time goes by, the probability a given option will move into the money decreases, and so the theoretical liability of the seller decreases accordingly.
My Experience Adapting:
Having come from a background in a long vol fund, malleability has been essential for me to revise my beliefs around shorting vol. Principally, being a reader of Taleb, I am still rather reluctant to having permanent tail-risk and so, I look to hedge such outcomes when structuring a trade to limit worst-case drawdowns to something pre-defined. Additionally, I am reluctant to enter the short vol side without having a view on future implied volatilities. To satisfy these preferences, shorting vol after a meaningful expansion in implied vols is when I feel most comfortable… Perhaps because of that elevated premium!
In 2017, this worked favourably in a couple of instances, both were geopolitical in nature, where political catalysts caused outsized realised vol due to markets overreacting — and implied followed accordingly.
The Anatomy of Geopolitical (Over)Reactions – A Behavioural Model:
Distinguishing between genuine geopolitical events where there is a digital repricing and those which are simply a liquidation or de-risking event — with a later re-risking — is no hard science. In either case though, realised vols have already expanded and implieds naturally chase in response. Thus, selling a strangle is of worthy consideration, especially in the liquid foreign exchange markets where double digit moves only tend to occur when somebody abandons a fixed currency regime (cough… SNB).
Polemic Paine (@polemicpaine) one of finance twitter’s resident gentlemen and one of its more generous contributors, pointed out a brilliant model in his post Markets. Where Physics Meets Psychology which fairly accurately depicts the mental model one must apply for such events. It is called the damped harmonic oscillation. With time, the amplitude of the oscillation decreases, just like the pitch of investor sentiment as market participants calibrate to the shock.
IMAGE 1: The Damped Harmonic Oscillation
Evidently, it seems in the examples I shall outline below, this framework held true. Behaviourally speaking, this is because the shock value decreases with each newer instalment of the same story — much like how “Grexit” was seemingly of much greater dramatic magnitude after the first Greek default, than the later n reincarnations of the same disaster.
Additionally, there’s a mechanistic reason for this process too, in that, after all the leveraged weak hands are washed out, all the positioning left is in it for the long haul. Accordingly, these less fragile participants, with longer temporal horizons inadvertently reduce the probability of high realised volatility from stop runs and news events.
Case Study 1: MXN on Trump’s Election
As a function of Trump’s rhetoric regarding the border wall, the potential renegotiation of NAFTA and other fairly protectionist views within the Donald’s policy platform; the Mexican Peso traded like a real-time referendum of speculators expectations of the election. Polling was shown to be rather inaccurate, perhaps because of the social stigma associated with admitting to being a Trump voter. This was further compounded by betting markets which had priced in a Clinton win, in a curious one-mimics-other information feedback loop. Thus, the November 8th election of Trump corresponded to an almost eleven sigma rate of change for a singular daily observation in the Mexican peso, as evidenced below.
CHART 1: USDMXN Daily Rate of Change with linear regression std dev +4 (red) +2 (red), -2 (green), -4 (green).
Further, the 30 day realised volatility of the currency leapt to a near 3 sigma deviation to historical vols.
CHART 2: USDMXN (white bars) vs USDMXN 30d realised volatility (blue) with linear regression std dev (+3, -1)
Interestingly, whilst the move was so digital in nature, implieds remained elevated whilst realised volatility managed to fall to below it’s 5 year mean (the green line above), bottoming near 11 before the end of 2016.
Suppose one had waded into the chaos and sold a 1m strangle to exploit the elevated risk premium attributed to the Mexican peso because of Trump’s election, it’d have been a rather nice period of positive theta to see out the end of the calendar year… Whilst some less than experienced commentators panic sold all of their equity holdings!
Whilst I didn’t have the vision to partake in the normalisation of the Peso after the Trump-shock, it did reinforce my earlier experience from 2016’s other major supposed political shock…
Case Study 2: GBP on Brexit
On the 23rd of June 2016 the people of Britain voted against remaining in the European Union. Meanwhile, polling — as it did later in the Trump polling debacle — suggested the Remain camp had it. Presumably, the bookies were watching this polling and thus they were completely discounting the outcome of Brexit. While the market was watching these two derivatives, instead of options pricing to gauge the likely outcome… Arguably, a less reliable process than having a pint at a pub with the locals anywhere outside of west London!
This all amounted to a rather significant political ‘shock’ with a 14 sigma daily rate of change in the cable (below) and an 11 sigma move in EURGBP.
CHART 3: GBPUSD Daily Rate of Change with linear regression
Whilst I didn’t take action on the short vol side after Brexit, I did attempt to buy an early dip and was stopped out. Arguably, I was applying the wrong framework. I did think Brexit was a coin toss of a chance, but had misunderstood the asymmetry it represented – a worthy topic for a later essay. As we can see below, it wouldn’t have been an unreasonable approach to short vol once the July low was in place, as it was over a quarter of sideways action before the eventual re-test and break of the low.
CHART 4: GBPUSD (white bars) vs GBPUSD 30d realised volatility (blue)
Presumably, these observations are what later encouraged the framework to ‘click’ as I later wrote in my February 2017 piece The Common Law Premium – Buying the Pound Sterling:
These implied volatility levels are invariably elevated as a function of the grey swan-like impact of Brexit. Arguably, they price a measure of recency bias given the likelihood of another ~14 sigma daily move is low, which fits with my unpretentious observation that markets ascribe a greater risk premium to political uncertainty than to its antithesis.
This view justifies the perspective that both implied volatility and realised volatility will revert toward the mean as markets re-calibrate to a more certain political reality, in a market where the weak hands have been shaken out due to such high realised volatility.
Simply put, less fragile positions reduce the probability of high realised volatility. However, this is not to suggest exposure to unlimited loss structures is ever prudent. Instead, loss-limited structures, such as short put spreads; enable the ability to collect this elevated risk premium with a pre-defined and limited loss. Similarly, for those who trade esoteric structures, selling one-touch puts provides a similar return profile to put spreads as they are loss-limited, non-recourse structures.
In the piece, I was referring to GBPUSD 1m implied volatility still being in the 89th percentile of its historical distribution.
CHART 5: GBPUSD 1m Implied Vol as at February 2nd 2017
At various times this year I exploited this elevated premium, originally short puts and on a couple of occasions later, short strangles. Nevertheless, specifics aside, this gave me the experience to quickly identify another opportunity later in the year.
Case Study 3: EWZ on the Temer Revelations
On the 18th of May 2017, EWZ gapped down nearly 20% on news of recordings of Brasil’s President Michel Temer supposedly making bribes. Presumably, the angry crowds on the street and congress members demanding his impeachment contributed to a slightly over 8 sigma move in the rate of change of the EWZ index.
CHART 6: EWZ Rate of Change with linear regression
To the best of my recollection, having mulled on it over the weekend, I thought there was some counter-intuitive logic that, even if Brasil impeached its second President in a row, after Dilma Rouseff, it signaled an unwavering dedication to enforcing the rule of law and upholding the people’s will… A bullish sign!
As a result, I shorted some June and July expiry structures which contributed a collective 420bps to the portfolio over that period.
Short Vol Alert Model:
Going forward, I have developed a couple of models in TradingView which will alert statistical outlier moves in realised vol in particular markets to encourage one to assess whether they fit the framework… Indeed, deciding whether to follow a signal and the appropriate method of expression requires a lot of consideration. Nevertheless, these alerts going forward will presumably allow one to hopefully exploit the dynamic outlined herein and share these trades with the Prometheus Macro Premo Social members.
CHART 7: Statistical Short Vol Model 1
CHART 8: Statistical Short Vol Model 2
Conclusion:
Whilst it’s difficult to know for certain if this approach is superior to continually selling vol, it strikes me as shifting the already attractive win/loss ratio of short vol a touch into one’s favour for the reasons outlined above… Further, it seemingly would reduce the probability of a black swan type event marking implied vols against you in a manner which is terminal for one’s portfolio. Meanwhile — extending the biological analogy further — it largely removes the eventual malignancy of exposure to unlimited loss from constantly shorting vol in an infinite time series.
With that said, I welcome any feedback from you all via the AMA or the comments section below.
Best,
Carl Hodson-Thomas
Subscribe: To see my trades and ideas in real-time, sign up to follow me via Premo Social.
