1 october 2020
Quantitative robustness, tailor-made to your portfolio
When we ask investors whether they would be willing to introduce quantitative tools into their investment processes, they usually embrace them enthusiastically. It is now consensus that quantitative approaches bring robustness, diversification, risk management and discipline to investment decisions, generally leading to better outcomes.
Still, a relatively small number of them actually manage to make the transition. Rather, they stick with previous frameworks, generally implying a strategic reference allocation as well as tactical deviations. Unfortunately, such approaches are vulnerable to emotional biases, evolving volatility regimes and changing correlation patterns.
Two key reasons make the implementation of quantamental processes non-trivial in real life. First, the simplest optimisation methods yield unrealistic outputs. Mean-variance methods, such as Markowitz optimisations, tend to result in highly concentrated portfolios that saturate assets with the highest Sharpe ratios. This implies extreme sensitivity to the inputs, which by nature are uncertain and in turn, limited robustness when those inputs change. Investors typically need to overcome these drawbacks by imposing a large number of constraints on the optimisation, resulting in a high degree of discretion and judgment. Unfortunately, this is exactly what quantitative methods are trying to minimise.
The second important shortcoming to Markowitz methods is the necessity of coming up with numerical tactical views. Mean-variance optimisation requires to form convictions on every asset in the investable universe and express them in the form of a precise number that will be used as an expected return. Unfortunately, this approach is a poor match to how market convictions are usually formed: investors tend to have incomplete views, most often expressed in relative terms, and marred by uncertainty.
At SILEX, we believe that quantamental investing is the future and should be available to everyone. The ability to combine expert human convictions with quantitative tools brings invaluable benefits to portfolio construction and, ultimately, performance.
But quantamental investing can only be widely used if made simple. Our SPARK platform brings together a set of tools that allows investors to harvest the benefits of robust portfolio optimisation with a high degree of customisation.
SPARK Allocator makes portfolio optimisation unprecedently easy. The tool follows the Black-Litterman approach, a widely used optimisation method that overcomes many of Markowitz’ issues. Initially introduced in the early 1990s, the approach uses Bayesian calculus that “blends” equilibrium expected returns derived from the strategic portfolio with an investor’s tactical views about the market. In other words, Black-Litterman optimisation brings quantitative rigour to the intuitive process of overweighting or underweighting asset classes in a discretionary manner.
Computational experience has shown that portfolios constructed through this method are more stable and better diversified than those constructed from the conventional mean-variance approach. This is because tactical views are considered for what they really are: incomplete, uncertain, relative bets. The BL optimisation is mixing long-term equilibrium returns that are implicit in the strategic reference portfolio with tactical views that can be expressed as simple relative views with various degrees of conviction.