This time of year one often encounters certain football-fans’ truisms. Sometimes one is told “the best defense is a good offense.” At other times, “the best offense is a good defense.” At yet other times, “it’s all in the special teams.”
The gist of them all is: the distinction between point-scoring/alpha-seeking on the one hand and risk management on the other can be overdone. They both have to make up one team.
I thought of these maxims while reading Cristian Homescu’s take on the issue of “tail risk.”
Swans or Turkeys
He starts with the issue of the “black swan” versus the “black turkey.” As regular readers of AllAboutAlpha know (and if you are new to this blog or the subject, welcome!) a black swan is a statistically unlikely event, generally one more than two standard deviations from the norm; something that could not reasonably have been foreseen, but that happens anyway, with calamitous consequences.
Was the financial crisis of 2007-08 an example of a black swan event? Some of those who have argued against that proposition, who have argued that it was foreseeable – it simply was not foreseen due to willful blindness – have described their own view as the “black turkey” theory.
Homescu leans toward the black-turkey view, but his bottom line is that the importance of mitigating tail risk is great on either reading. For the incidents of willful blindness in the past suggest there will be willful blindness in the future, so the turkeys are as dangerous as the swans unless the risk they present is managed.
Homescu then quickly runs through many of the common issues is the literature as if offering a tour: metrics for tail risk, extreme value theory, the various fat tailed distributions (student t, hyperbolic, generalized hyperbolic, etc.). He slows down a bit when he comes to the matter of “regime switching.”
A regime switching model may treat a high-volatility environment as one “regime,” and a low-vol environment as its successor regime. The idea, as it applies to risk management, then, is simply to be ready in either setting for the switch to the other. One manifestation of this is the “turbulent time indicator,” which Homescu describes as “an out-of-sample indicator built using macroeconomic data and relying on a Markov switching model to determine states of calm market, turbulent bullish market, and turbulent bearish market.”
In a recent paper, the scholars behind the turbulent-time-indicator (Hauptmann et al.) claimed that their three-regime approach combined with a mix of macroeconomic factors amounted to an “early warning system to timely forecast turbulence in the U.S. stock market.”
Homescu’s favored approach to tail-risk management, the patient reader eventually learns, is one that relies on five “regimes,” rather than the mere three of the Hauptmann paper. These regimes are: crisis; high risk aversion; normal; low risk aversion; euphoria.
The point he really wants to make, though, the point of the tour, is in his mind closely connected with regime schemata. He holds that certain tail-risk management strategies do more than just manage risk. They make money. The distinction between defense and offense is relative.
A Five Regime Approach
The five-regime model, which Homescu takes from the work earlier this year of Frederic Dodard and Robert Benson uses VIX futures as a key risk management tool. Specifically, it provides that traders “go long VIX futures when the MRI [market regime indicator] crosses from a Normal to High Risk Regime.” They should sell, that is, close out those positions, in one of two scenarios.
The first such scenario is that the MRI crosses into “Crisis” territory, and then falls back into High Risk. If that happens, then closing the VIX long position “is analogous to using a trailing stop on a profitable trade.”
The second scenario is that the MRI could fail to get to Crisis at all, but could fall from High Risk back to Normal.
A Starting Point
In both instances, a trader’s use of the portion of the portfolio he allocates to VIX futures will take advantage of the mean-reversion characteristic of the VIX at its extremes. Let profits run on a VIX futures position, allow it to make new highs, but close out at some as some weakness sets it.
As the graph above shows, back-testing this model against market conditions since the end of 1998, Dodard and Benson contend that the use of short-term VIX Futures would have added 285 basis points to annual return. The results from the use of medium-term VIX futures would also have been positive, though less dramatic.
Homescu concludes that this, and kindred strategies “may serve as a starting point for investors and asset managers interested in minimizing the tail risk of their portfolios.”