Working with pre-defined rules, systematic investing requires at the outset that we understand a portfolio's sources of risk and reward. To make a process robust and sustainable, a strategy should not depend excessively on market parameters or past price behavior. A degree of freedom is best left to take care of unseen markets or events. Allowing for this will make the strategy more effective in dealing with risks. The problem is that the outcome of this action cannot be quantified from past data. The alternative of buying an expensive hedge to cover all scenarios is not viable either.
Adapting to the unknown
A different perspective on risk can be helpful in getting to a more effective solution to deal with this unknown.
We believe it is better to separate (i) “average,” or “normalized” risk, from one that is (ii) unpredictable, random and of a “tail” nature. Average risk, to be clear, arises from the normal operation of economic systems leading to changes of asset prices over time. While also used to price the risk premium of an asset, it represents the asset's volatility as demand and supply conditions look to find its fundamental valuation. Put another way, it is the uncertainty premium around the natural process of market adjustments.
Unpredictable or tail risk, on the other hand, cannot be averaged and therefore is different. It carries an uncertainty premium that is (i) large in magnitude, (ii) infrequent, offering no predictability on the timing, and is (iii) relatively short in duration. It tends to be orthogonal and therefore mutually independent of the factors that ordinarily determine the average risk behind an asset.
Worth noting is that a tail event can influence the average risk in asset pricing over time, depending on the nature and magnitude of the impact. Tail risk could be seen as the uncertainty around a violent tsunami that leaves in its wake a trail of destruction. The shock is usually unpredictable, short-lived, devastating in impact and random in timing. The global financial crisis of 2008 created havoc and led to the Great Recession, which we are still struggling to get out of.
When crafting a strategy it is therefore worth considering (i) how we use average risk for generating long-term reward; and quite separately (ii) how we deal with tail risk. Separating the actions for each is likely to make the strategy more robust. Mixing them dilutes power and can lead to a loss of skew or profit potential in the strategy.
In a quantitative model, one can unknowingly penalize overall portfolio returns by overestimating large tail risk compared to lower average risk if both are treated in the same way. It is much like carrying an umbrella whether it be on a bright day or one that is cloudy, just in case it rains. The tail risk event that sees a market plummet unexpectedly is distinctly different from an average lowering of the mean leading to a short trade.
Fundamentals can result in downward drift in a market, thus taking a position short. This would be the result of an average risk adjustment, not a tail risk event. To deal with a tail is to buy insurance. One assumes there is little or no predictability in the loss from a potential calamity both in magnitude and timing, and yet it is needed. Tail protection is therefore better dealt with in a different way. Directional strategies such as commodity trading advisers and macros often offer optionality in crisis periods, not by design but as an output. However, this crisis alpha is not to be taken for granted. It is hence more effective to bring this insurance into the strategy as a separate part.
When designing a risk strategy, it pays to allow for a reasonable margin of error. This buffer for being wrong increases the strategy's robustness and makes it less fitted to past conditions. There is less straight-jacketing and more room to maneuver. A narrow margin of error can result in mispricing both tail and average risk. A wider margin of error in setting parameters generally helps in facing new challenges.
Risk allocation methodologies for a portfolio generally take two directions: static or dynamic. The static approach relies mainly on individual market timing for returns. The methodology silos risk capital by market. Average risk measure is used to decide on capital allocated to a market and, when desired, market correlations also are considered. With such methodology, long run volatility and correlations are expected to hold and market-timing becomes the driving force for returns. The sum total of the individual market rewards gives the total portfolio alpha and the approach is somewhat linear.
The second is a dynamic approach. Such an allocation methodology generates alpha from shifting capital between various risk factors. The process has a greater portfolio feel. Such dynamic allocations can adapt to fundamental changes in asset volatility and correlations. They are less dependent on market timing, placing greater emphasis on the portfolio, and carry a wider margin of error. Relative positioning of one market against others in the portfolio drives the alpha rather than market timing.
While the choice between the two styles is down to individual preference, the dynamic approach is likely more robust. Designed effectively, it not only can avoid dependency on market timing to generate main portfolio returns, but also reduce cost of errors from noise. Using average risk in considering the relationship of one asset with another, the dynamic approach is better equipped to align the portfolio to changing macroeconomic themes. While market direction is not unique to each macro theme, the proportion of the portfolio risk allocated is different and rarely the same. As such, if say a deflationary environment is brought by cyclical low growth and demand, the portfolio positioning can be different from that which comes from supply-side fixes such as those seen from monetary authorities in recent years.
In conclusion, a dynamic risk allocation process that is able to separate average risk from tail risk without relying on predictability from past data, is likely to yield more reliable results in the long run. It can be fundamentally intuitive and yet offer the rigor of systematic execution. In this context, the greater the separation of the two risks, the better is the ability of the strategy to perform as environments evolve. Action for dealing with tail risk, in our opinion, is greatly enhanced when brought in orthogonal to the average risk of the portfolio, much like the separate purchase of an insurance. With a wider margin of error, this holistic approach prepares a systematic strategy to gain from most market environments including dealing with those rare but devastating market tsunamis.
Aref Karim is CEO and chief investment officer of Quality Capital Management Ltd., London.