By Martyn Wild.

We've had such a positive reaction to our recent article, 'Who wants to be normal, anyway?', that we thought we'd produce a follow-up piece. Readers will recall that in that blog, we shared our opinion that contemporary asset allocators may not be using the most up-to-date techniques to construct portfolios. Worryingly, we found that the common type of analysis used to assess the likelihood of portfolio outcomes was liable to fail clients when they needed it most: in times of extreme negative returns.
Our solution is to encourage the use of non-parametric analysis, which we believe is far better at estimating the probability pattern of market returns and as a consequence, is more valuable as a foundation for determining the portfolio that is ultimately selected for a client.
Two trees does not a wood make
The famous idiom, '...you can't see the wood for the trees' is a good way to characterise a common weakness in contemporary portfolio construction; an obsession with just risk and return. Using risk and return as part of your selection methodology isn't a terrible idea as far as it goes but is somewhat, well, two dimensional. Portfolios are complex creatures that cannot be reliably defined by two characteristics - especially if, like us, you believe that returns are non-normally distributed. Moreover, there is almost always more than one portfolio that can meet the investment objectives, so its important that you apply a robust selection mechanism to zone in on your preferred structure. So while risk and return should absolutely form part of the selection criteria, don't stop there.
A wood is comprised of many different types of tree
Generally speaking, asset allocators determine an efficient frontier - a short-list of candidate portfolios with the highest return for any given level of volatility - and then select the portfolio based upon their investment philosophy. Sometimes, this is the portfolio with the highest Sharpe Ratio. Often, it's the one with the highest risk (and therefore highest return expectation) they can stomach! There are probably as many techniques as there are portfolios...
Chart One - An Efficient Frontier
In Chart One, we illustrate an example efficient frontier. We also include the portfolio defined as optimal from a Markowitz perspective, i.e. the highest Sharpe Ratio (red dot).

As for our particular approach, don't be surprised that we aren't about to disclose our proprietary portfolio construction methodology! Nevertheless, what we can say is that in addition to the usual suspects, we like to utilise metrics like conditional-value-at-risk (CVAR)*, given the non-normality of market returns. CVAR measures the average expected loss in extreme scenarios or equivalently, assuming you need to liquidate your portfolio each time markets go to pot, how much of your capital value you could expect to lose. To recap our prior article, assuming normality will likely understate CVAR for high volatility portfolios and overstate it for low volatility portfolios. Therefore, constructing portfolios including CVAR as a criteria is more likely to ensure your portfolio is 'all weather'. We are similarly captivated with each portfolio's likelihood of achieving its key objectives.
The question of how we then weight each portfolio metric is perhaps the more interesting question to answer. You see, at the risk of sounding deliberately vague, it depends! Why? Well, let's use the investment horizon to illustrate our point.
How many trees of each type in the wood?
Portfolios with short-term investment horizons (generally 5 years or less) - particularly those with a high allocation to equities or other risk assets - are very susceptible to extreme negative events. As such, we would bias our weighting of metrics like CVAR over expected return every day of the week. The corollary is also true: for long term investors (think at least 10 year investment horizons), we would bias the weightings more toward returns than CVAR given the effect of temporal diversification over long time horizons. Refer to our article 'Take it Easy' for a fuller description of what we mean by temporal diversification. What doesn't change are the metrics we consider. What does change is how we apply them.
Something that may also surprise you...our weightings are also influenced by factors like the client's unique utility function (or in plain English, their preferences) and often, the relative performance of peers. What is for certain though is that portfolios (and therefore the weighting of component metrics) are unique to each client and cannot be defined by risk and return alone. Sometimes the highest Sharpe Ratio portfolio and the one we choose are one in the same, but we think that's largely coincidental and a product of our pragmatic bias toward return efficiency. Invariably, they differ when the needs of the client and true nature of markets are taken into account.
The knack, we find, is to take a step back and look at the wood, not the trees...
For more information or to talk to a member of the MARQAM team, please click here.
MARQAM is a privately-owned, boutique consulting company focused on providing superior investment outcomes for clients and greater profitability for businesses. We are not affiliated with any other financial institution
Disclaimer: The information provided here is for interest purposes only and does not constitute investment advice or a recommendation of any kind.
*Our clear obsession with using CVAR in portfolio construction is a function of the asymmetric sensitivity of clients to losing money as opposed to making money and because market returns are usually not normally distributed. CVAR is an enhancement on VAR and gives a more useful measure of the likelihood of extreme negative events, such as the GFC or Tech Wreck.
Comments