• The “confidence heuristic” is a mental shortcut defined by behavioral psychologists as the tendency to believe that confidently expressed arguments signal better information.1
• Traditional mean-variance portfolio optimization has been criticized as “estimate error maximization” because it places one-hundred percent confidence in its estimated risk and return inputs. Attempts to build uncertainty into portfolios have been well-researched, advancing the art of modern portfolio construction.
• In times of regime change in markets, backward-looking portfolio construction tends to suffer as assets behave in ways that do not reflect the past. Implementation is as important as theory when building actively managed modern portfolios.
In Part One of this two-part Two Minute Tuesday series, we discussed the current lack of confidence shown by market and economic participants, and how extremes in confidence can lead to self-reflexive markets in which prices become detached from fundamentals. In Part Two, we examine the uses and (misuses) of confidence in building portfolios.
Beware the Confidence Heuristic
The “confidence heuristic” is defined as the tendency to apply confidence as a means to predict accuracy. Most media pundits and strategists are aware of this heuristic and leverage it to engage their audience.
A problem arises when one has to actually build portfolios, because investment decision-making is an environment of control with no certainty. Most investors buy or sell based on the outcome that they imagine, and their imagination is a function of the investor’s level of confidence. The degree to which an investor is certain of an outcome is an expression not of the probabilities of that outcome but the investor’s own level of confidence. The best investors are far more ambivalent about the outcome they imagine. They consider the absolute opposite scenario and appreciate the less they are able to consider that scenario the more likely it is to happen, because their belief system is too strong in a world of imagined outcomes.
Be the Fox, Not the Hedgehog
Dr. Philip Tetlock, one of the authors of Superforecasting: The Art and Science of Prediction, has been conducting experiments on human judgement for the last several decades. Findings from his studies led him to make a distinction between “foxes” and “hedgehogs”, a metaphor borrowed from ancient Greek poetry and popularized by the philosopher Isaiah Berlin: “The fox knows many things but the hedgehog knows one big thing.”2 Hedgehogs tend to be more confident and more likely to gain media attention but, as Tetlock’s research finds, they also tend to be worse forecasters. In contrast, foxes tend to think in terms of “however” and “on the other hand”, are able switch mental gears and talk about probabilities rather than certainties.
Such distinction can also be made in the art of portfolio construction where forecasting is a critical input, and in particular the foundation of modern portfolio theory, where a group of assets’ “efficient frontier” can quickly rotate when one considers the unavoidable estimation error in expected returns. While the theory of mean-variance optimization (MVO) was revolutionary, one of its criticisms is that it considers the inputs (the estimates of assets’ return and risk) as certain, which creates the tendency to maximize the estimates’ errors, leading to unintuitive weights and highly concentrated portfolios.
The remedy chosen by most institutional investors is to explicitly constrain the optimization to “preferred” weight ranges, significantly reducing the solution space for the optimizer to search. An alternative option is to build uncertainty into the portfolio construction process explicitly.
Embracing Uncertainty, Updating Continually
Several academics and practitioners have researched the incorporation of uncertainty, or estimation error, explicitly into modern portfolios. One of the more famous examples is the Black-Litterman model, created by Fischer Black and Robert Litterman, which uses a Bayesian process to blend market equilibrium returns with the subjective views of an investor.3 Depending on confidence parameters placed on each investment view, the Bayesian process places less weight on the point estimates of return of the investor and helps the optimization to overcome the problems of unintuitive, highly-concentrated portfolios often observed in MVO.
To illustrate, below is an example of the eleven Global Industry Classification Standard (GICS) sectors of the S&P 500 Index whereby we have set investor views to prefer Communication Services, Energy, Health Care and Utilities over Consumer Discretionary, Information Technology and Financials. To simplify, we parameterize the views such that the outperforming sectors are expected to achieve 2% annualized excess return over the underperforming sectors.
As shown in the chart nearby, 0% Confidence on the views leaves us with the market benchmark portfolio. As confidence in our views increases, the relative weights of preferred sectors increases while that of the less preferred sectors decreases. As we near 100% Confidence in our views (which is the default setting for MVO) we end up with a portfolio of only our preferred sectors. By engaging a Bayesian process that starts with the benchmark portfolio and slowly adds confidence to the views, we’re able to engineer an optimized portfolio that is less concentrated, more diversified and still expresses our views.
A Time for Active Management
For the first time in a while investors are confronted with a regime shift across markets that favors the active manager. The previous 13 years of excessive monetary policy have generally favored a lower-for-longer approach and created an environment which heavily favored passive investing. During these times of easier monetary policy and ever-expanding margins, broad equity market performance was easier to get right as businesses could access capital at record low rates regardless of balance sheet strength or earnings power.
In an environment of persistent inflationary pressures and rising borrowing costs, fundamentals matter again. Even so, such an environment demands the investor be aware that the range of outcomes for geopolitical tension, the yield curve, company earnings and credit risk are wider than we’ve seen in a long time. With multiple stacked vulnerabilities afflicting the economy and markets, now is the time to be able to switch mental gears and think in terms of probabilities rather than certainties.
By incorporating uncertainty directly into the portfolio construction process, investors may have an opportunity to avoid overconfidence in their portfolio construction process and build more diversified and robust portfolios for their end clients.
Important Disclosures & Definitions
1 Jonathan P. Thomas, Ruth G. McFadyen, The confidence heuristic: A game-theoretic analysis, Journal of Economic Psychology, Volume 16, Issue 1, 1995, Pages 97-113, ISSN 0167-4870, https://doi.org/10.1016/0167-4870(94)00032-6.
2 Berlin, I. 1953. The Hedgehog and the Fox: An Essay on Tolstoy's View of History. University of California: Weidenfeld & Nicolson.
3 Black, F. and Litterman R. (1992) Global Portfolio Optimization. Financial Analysts Journal, 48, 28-43.
Performance data quoted represents past performance. Past performance is no guarantee of future results; current performance may be higher or lower than performance quoted.
S&P 500 Index: widely regarded as the best single gauge of large-cap US equities. The index includes 500 leading companies and covers approximately 80% of available market capitalization. One may not invest directly in an index.
AAAI000183 08/31/2023
