Most advisers and their clients would agree that a key aim of investing is that over time, returns will beat the rise in the cost of living (after fees and taxes).
In addition, it is important to try to minimise the risk of permanent capital loss. This isn’t mark-to-market losses driven by sentiment and noise, but losses that arise from a permanent diminution of business value.
Different investment managers will seek to do this in different ways, depending on their investment philosophy and style as well as their investment area or asset class.
At Quay, we seek the best real estate investment opportunities that can meet our investment objective with the lowest possible long-term risk. This may sound simple, but it usually isn’t. There are a number of elements that go into the construction of a portfolio. Here we outline our approach to building a high conviction, index agnostic fund, and seek to answer common questions such as: beyond selecting stocks, how do we allocate weights? And how many stocks are too many (or too few)?
The first step for an investment manager is usually to define the number, or range, of securities they are prepared to own in their fund.
Portfolio construction often begins with the best investment idea – so the portfolio comprises of one security with 100% weight.
Adding a second security offers some diversification, but also dilutes the best idea. However, the known benefit of diversification tends to outweigh the risk of only investing in one idea. The same can be argued for adding a third, fourth, fifth security.
So there is a cost benefit for every stock that is added to the portfolio. But at what point does the benefit of diversification begin to outweigh the cost of diluting the best ideas?
At Quay, we sought to define the ‘sweet spot’ of diversification versus dilution by calculating and recording the standard deviation of returns of 10 different portfolios of two randomly listed real estate securities. We then repeated the process for three securities, then four, and so on, up to a portfolio of 90 randomly-selected securities. In total, we recorded the average volatility of 890 portfolios in total, plotted in Chart 1.
By applying a ‘line of best fit’ to the data, we identified a clear pay-off diagram between measured risk on the y-axis (volatility) and dilution measured along the x-axis (number of securities). The average difference in volatility between a two and a 90-stock portfolio is greater than 2% standard deviation points per annum. However, around half the volatility reduction occurred after just 20 stocks, and three quarters of the volatility is reduced with 40 securities.
Beyond 40 securities, the benefits of additional diversification are negligible – yet the cost (dilution) for additional securities is roughly the same. At the other end of the spectrum, fewer than 20 securities begins to add meaningful volatility and we believe that for retail investors, the risk of fewer than 20 securities is too great.
Within this context, we concluded a portfolio of between 20 to 40 securities was optimal for our strategy. For other asset classes, the numbers may differ but the logic is the same.
When working out the individual stock and sector weights, a few elements need to be factored in.
Chart 2 maps the expected return distribution between two securities, ‘A’ and ‘B’. Both securities are expected to exceed the real return objective of 5% per annum. But while B displays a higher expected return as measured by the median (8.5%), we would tend to favour and assign a greater weight to A (expected return 6.5%). This is because we are more focused on the left hand side of the chart (the risk of total returns falling below our investment objective) than the right (the risk of outsized returns).
In this example, the cumulative probability that B’s returns could fall below our benchmark is 36.7%, while for A it is just 11.2%.
Unfortunately, this type of analysis is prospective in nature. There is no statistical data that allows us to map the variance of returns as neatly as the chart above. However, through our fundamental research, detailed cashflow forecasts, sensitivity analysis and industry analysis we can form a view on the variability in our expected returns.
High conviction ideas must not only meet our return objective with a margin of safety (and have an assessed low distribution of returns), but also have another characteristic – a skewed or asymmetric pay-off.
Chart 3 includes security ‘C’, where the returns are skewed to the upside and the downside is limited. Of the three opportunities, C would represent our highest conviction idea and therefore the highest portfolio weight.
An example of these asymmetric returns could be a takeover offer, where the underlying security price is supported by a cash bid but there is the prospect of a higher offer.
The downside risk is usually limited to the bid price, depending on the terms of the offer. But once a bid is public these are easy games to identify, so the upside is limited.
Longer-term, these scenarios also exist but are less obvious. In listed real estate investing, buying assets at a substantial discount to replacement cost – or assets that are supply constrained and have a substantial barrier to entry or secular long run tailwinds (such as demographics) – also has very limited downside and skewed upside potential, but only if viewed through the prism of a long-term investment horizon.
Of course, each asset manager has their own approach to portfolio construction. But understanding the thinking behind the selection of securities can help advisers identify the opportunities that best suit each client’s needs.
We will always favour lower (acceptable) total returns with lower downside risk, rather than chase the ‘hot sectors’ of today that may promise higher returns but come with meaningful downside risk.
Chris Bedingfield is principal and portfolio manager at Quay Global Investors.