The fundamental law of active management in context: discretionary vs. systematic?

Written by Sabrina Herold at Tom Capital AG

Researchers comparing active management approaches often try to portray one as superior to the other, but is this necessary and even wise? What framework might be useful to review both approaches as an investor, and what are the underlying elements?


In this article, we tackle these questions and more by taking a fresh look at systematic and discretionary investment styles utilizing the framework of the fundamental law of active management.


We describe the term discretionary as decided according to human judgment; not decided by rules. In contrast, systematic shall be defined here forth as decided according to rules of a system that has been designed by one or more humans.


The fundamental law of active management originally mentioned in the paper by R. Grinold and R. Kahn in 1989 describes the productivity of a fund manager as determined by his/her skill or Information Coefficient (IC) in forecasting exceptional returns and the breadth as the frequency at which this skill is applied, measured in the number of independent trades. The consequence of these two elements describes the value added to the portfolio strategy, or Information Ratio (IR).


In mathematical terms, we specify the necessity of skill and breadth to add value as:


This equation (1) denotes, that each unit of skill has a relatively higher impact on value-added and that each unit of additional breadth creates diminishing marginal returns.


Skill in forecasting exceptional returns

In a discretionary investment approach, most investors consciously or unconsciously judge skill based on each individual trading decision executed by the manager. Portfolio construction and risk metrics also play a role, but often and unfortunately to a lesser degree in the initial assessment. Hard-to-measure qualitative criteria are also considered to assess skill, some of which may simply boil down to personal liking.


In a systematic investment approach, the ability of an entire system or model, including risk management and portfolio construction, to outperform the market is generally consulted as the measure of skill. The people building this system are the discretionary component of the systematic skill and are usually studied to a lesser extent. In our view, team composition, experience, and skill set of core team members deserve more attention to estimate the long-term success of any systematic investment.


The argument in favor of the skill of discretionary models is that one may say, in unprecedented market times, a discretionary manager is on average more likely to prove higher skill in the short term, being able to react quickly to the newly established rules. A systematic approach would need to wait until a shift in market behavior becomes visible in the data for the model to change directions, possibly realizing a performance drag. However, this benefit balances out in the long term, depending on the multitude of studies and points of reference taken.


A qualitative factor frequently attributed to strategy skill is the transparency each investment approach offers. Systematic managers are often accused of working incomprehensibly in a "black box," while discretionary approaches are said to be the opposite. Looking at the systems each of them utilizes for decision-making, one could claim that the opposite is true.


Discretionary managers make decisions in their human system. Despite the common belief of the homo economicus, human decision-making has been proven to be biased and irrational by Nobel Prize winners, such as Daniel Kahneman and Vernon Smith (2002), Robert Shiller (2013), and Richard Thaler (2017).


For systematic managers who outsource their investment decisions entirely to a model, these biases are much less pronounced in the actual trading decision, although they are still present to some extend as machine learning models are developed and revised by humans. Even when it comes to testing the robustness of a forecast, systematic investment managers can rely on unbiased methods that are impossible for discretionary managers to use in making decisions, even when they use quantitative models as part of their investment decision (see our article referenced below for in-depth information).


Strategy breadth is measured by the number of independent trades

Technically, the most crucial determinant of strategy breadth is the capacity of the manager, team, or system to process information for trading decisions.


Discretionary managers are limited by the number of good quality assessments any singular manager and team of managers can make about a certain trade. Yet, discretionary managers do explore each trade in extensive depth, often unveiling indicators such as the management sentiment, which are harder to grasp today in numerical terms. Their inherently vertical nature allows discretionary investment approaches to focus the breadth of their strategy on a particular theme or niche.


In contrast, systematic managers are horizontal by design and diversify their portfolios across markets and often across asset classes. Their systems enable the screening of vast amounts of data across a broad spectrum in the shortest possible time to derive informed investment decisions. As the amount and quality of information reflected in the data continues to increase, much of the detailed information once reserved for discretionary managers is becoming accessible to systematic traders.


This transfer of knowledge is no one-way street from discretionary to systematic managers. Most, if not all, discretionary managers today build on advances in quantitative modeling and consult them for trading decisions. However, not to the extent that systematic managers do. A first study provides evidence, that combining both approaches too much could result in inferior outcomes (Grobys, Kolari, Niang, 2022). One rationale behind this might be that two coherent arguments never make one coherent argument when combined.


Value add to the portfolio strategy as a result of skill and breadth

Looking back at equation (1), we have shown the skill component to be measurable in both approaches, and we do not want to make any subjective conclusions about the superiority of one above the other.


Structurally speaking, the most crucial differentiator between both types of managers becomes breadth, with discretionary approaches generally providing fewer independent trades than systematic ones due to their vertical and respectively horizontal inclinations. As a result, risk metrics are generally lower for more diversified systematic instruments, whilst the return potential is usually stronger for discretionary instruments. Hence, investors can view systematic and discretionary investing as complementary, addressing different portfolio needs.


With strong heterogeneity across and within systematic and discretionary approaches, they can both bring about shining stars and foul apples to a similar extent. Examining their underlying skill and breadth in depth can help investors make the right investment decisions.


Tom Capital AG is an investment boutique that applies machine learning to the entire investment process, from data selection to portfolio construction. Do not hesitate to contact us at info@tomcapital.ch for more information or inquiries.


Sources


Richard C. Grinold, 1989, The fundamental law of active management, The Journal of Portfolio Research, DOI: 10.3905/jpm.1989.409211

Klaus Grobys, James W. Kolari, & Joachim Niang, 2022, Man versus machine: on artificial intelligence and hedge funds performance, Applied Economics, DOI: 10.1080/00036846.2022.203258

Campbell R. Harvey, Sandy Rattray, Andrew Sinclair, & Otto Van Hemert, 2019, Man vs. Machine: Comparing Discretionary and Systematic Hedge Fund Performance, Duke Innovation & Entrepreneurship Initiative, DOI: 10.2139/ssrn.2880641

Prequin, 2014, Discretionary vs. Systematic: Two Contrasting Hedge Fund Approaches, Hedge Fund Spotlight Volume 6, Issue 6, New York

Tom Capital AG, 2022, Testing the accuracy of your model’s predictions

111 views0 comments