Enterprising Investor
Practical analysis for investment professionals
31 January 2019

Ensemble Active Management (EAM): Taming Toxic Tails

Wide diversification is only required when investors do not understand what they are doing.” — Warren Buffett

What is Ensemble Active Management, and how can it help active managers outperform their benchmarks after fees?

Last September, we published “Ensemble Active Management: The Next Evolution in Investment Management,” the white paper that introduced Ensemble Active Management (EAM). In it, we explained how EAM portfolios are the result of the application of time-tested “Ensemble Methods” — core components of artificial intelligence — to the high-conviction stock selections of actively managed portfolios and that EAM portfolios’ superior performance is both repeatable and persistent. (For perspective, the 30,000 EAM portfolios outperformed the S&P 500 across rolling one-year periods 72% of the time, with an average annual excess return of 340 bps.)

We did not, however, include a comprehensive discussion of the time-tested investment principles that explain how — and why — EAM portfolios performed so well.

To be clear, EAM is not a strategy, an algorithm, or an overlay. It is an application of technology and creative problem solving to the means of building and delivering traditional investment portfolios. EAM portfolios solve a near-fatal flaw that is preventing traditional active managers from outperforming their benchmarks after fees.

Subscribe Button

The Active Manager’s Dilemma

One way for active managers to outperform their benchmarks is to focus on their high-conviction best ideas — what we call the Alpha Engine — and invest a substantial portion of the portfolio in them. This creates a Best-Idea-Centric portfolio. Such portfolios increase expected returns, according to academic research, and thus have higher outperformance potential. Some studies have focused on portfolio concentration to validate this concept, while others have built on the premise of Active Share as introduced by Martijn Cremers and Antti Petajisto.

While such a strategy makes intuitive sense and the research backs it up, managers rarely implement it because overconcentrating in Best Ideas triggers an unacceptably high risk of massive relative performance failures, or Toxic Tails.

How do we explain this paradox? Managers apply well-defined investment strategies to inform and drive their security selections, but by design, these strategies have embedded intentional biases that reflect the manager’s philosophy on how to achieve outperformance. Unfortunately, when a manager’s strategy is out of sync with market dynamics, there is a real risk for underperformance. When the portfolio is concentrated, that risk is magnified.

For example, if a manager believes that technology stocks will outperform, their portfolio will overweight the technology sector and over-allocate to tech stocks within their Alpha Engine. So if the technology sector lags the broader market, the manager will underperform and potentially trigger a Toxic Tail event.

As a consequence, managers have largely embraced a “safety first” approach, adhering to their version of the Hippocratic Oath: “First, do no harm (to investors).” The conventional response to Toxic Tail risk is to add a large number of stocks — which we term the Beta Anchor — to the portfolio. The purpose of these securities is more to manage risk and tracking error than to generate alpha.

However, there is a cost. Applying a full-sized Beta Anchor results in a heavily diluted Alpha Engine. In practice, it is like outfitting a long-distance runner with a helmet and pads before a race: It may reduce the risk of catastrophic injury, but it won’t help the runner outperform, let alone finish first.


Alpha Engine, Beta Anchor, and Best-Idea-Centric PortfolioAlpha Engine, Beta Anchor, and Best-Idea-Centric Portfolio


EAM portfolios take a different approach to managing portfolio risk. They embed a second layer of diversification at the investment strategy level, which reduces the risk of Toxic Tails such that the EAM portfolio can become truly Best-Idea-Centric.

The graphic below, based on proprietary research from Turing Technology Associates, demonstrates this dynamic. For this study, researchers evaluated the performance of 16 large-cap funds that Morningstar designated as Gold Rated in January 2017. By extracting each fund’s overweight positions relative to the S&P 500, renormalizing, and then updating every two weeks, they constructed 16 concentrated portfolios. The Gold Funds and the corresponding Concentrated Portfolios were compared over rolling one-year periods from March 2012 to September 2018.


Concentrated Portfolios and Increased Likelihood of Toxic Tails

Concentrated Portfolios and Increased Likelihood of Toxic Tails


The Gold Funds did well and delivered an average annual excess return of 0.9%. Thanks to the risk-management benefit of the Beta Anchor, the bottom 20% of the Gold Funds’ relative return distribution was limited to –6.8% and did not qualify as a Toxic Tail.

As research predicts, the Concentrated Portfolios outperformed with average annual excess returns of 4.2%. Unfortunately, their concentrated natures created an expanded negative tail, with the bottom 20% of the one-year returns creating a Toxic Tail of –1,040 basis points (bps).

Managing Toxic Tail Risk by Overdiversifying

As aspirational as a true Best-Idea-Centric portfolio might be, managers cannot justify that Toxic Tail risk. Thus, the conventional fund design features an oversized Beta Anchor instead of the optimal Best-Idea-Centric portfolio.

But the Beta Anchor dilutes the Alpha Engine’s net benefits and acts as a drag on returns. For example, assume a manager’s Alpha Engine delivers 200 bps in annual, undiluted excess return and that the fund has a Beta Anchor equal to 75% of the portfolio. The diluted, pre-fee result is thus 50 bps of annual excess return [200 bps × (100%–75%)]. After fees, the portfolio likely underperformed.

But if the Beta Anchor equaled 10% of the portfolio, then the pre-fee annual excess return would be 180 bps [200 bps × (100%–10%)], which leads to outperformance after fees.

The Ensemble Active Management Solution

So how do EAM portfolios overcome the structural design flaw that sabotages active managers? There are three steps:

  1. The Alpha Engine is extracted from multiple independent funds with unique strategies and biases. A full-Ensemble will use 10 or more funds, but a “mini-Ensemble” with just two or three funds generates tangible benefits.
  2. The Alpha Engines are combined through Ensemble Method techniques to create a multi-expert foundation for the final investment portfolio, thus introducing the second layer of diversification (at the investment strategy level) into the portfolio design.
  3. A new, enhanced Alpha Engine is built by integrating the underlying funds’ Alpha Engines and then inserting it into the new portfolio with little or no Beta Anchor.

Visual Depiction of Ensemble Active Management

Visual Depiction of Ensemble Active Management


The EAM value proposition comes down to the following two central outcomes.

1. EAM’s Added Layer of Diversification Reduces the Risk of Toxic Tails.

Diversification is one of the most critical tools available to portfolio managers. It decreases overall portfolio risk by reducing non-systematic risk and the dispersion of return distributions. Traditionally, diversification is introduced to portfolios at the security level. By injecting that second layer of diversification at the investment strategy level, EAM integrates multiple investment strategies and thus diversifies the individual managers’ biases and substantially reduces the potential for Toxic Tails.

To demonstrate, Turing built an EAM portfolio as part of the Gold Fund analysis discussed earlier. The graphics below illustrate the probability distribution of the Gold Funds, the Concentrated Portfolios, and the EAM portfolio, built from the 16 Gold Funds, over rolling one-year periods.


Probability of Relative Return Distributions

Probability Distributions of Rolling One-Year-Returns

Probability of Relative Return Distributions


  • Gold Funds had a high percentage (67%) of relative results around a band of +/– 500 bps and limited breadth of tails (width between dotted lines).
  • Concentrated Portfolios shifted results to the right, which indicates improved average returns, but had clearly expanded tails that spread from –30% to 40% (expanded breadth between dotted lines) and a significant exposure to Toxic Tails (5.1% rate) and suffered multiple “extreme” Toxic Tails of more than a –20% shortfall.
  • EAM portfolios dramatically compressed the tail distribution (reduced width between dotted lines), shifted results even farther to the right due to improved returns, with 88% in positive territory, and had no Toxic Tails.

The EAM portfolio’s added layer of diversification more effectively reduced negative tail risk than the industry standard use of a Beta Anchor.

2. Best-Idea-Centric Portfolios Increase Excess Returns.

But what about relative performance results as they relate to the Turing Gold Fund analysis? The performance table below expands on the first Turing chart, this time including the EAM portfolio results.


Improved Performance from EAM’s Best-Idea-Centric Portfolio Design
Rolling One-Year Returns, Net of Fee and Benchmark Returns

Improved Performance from EAM’s Best-Idea-Centric Portfolio Design


What are the key takeaways?

  • Gold Funds on average delivered an annual outperformance, after fees, of 0.9%.
  • The Concentrated Portfolios delivered excess returns of 420 bps, improving upon the Gold Funds’ results by 330 bps. Of course, this gain came at the cost of a significant risk of Toxic Tails.
  • The EAM portfolio achieved 550 bps in annual excess returns, capturing the 90 bps excess return of the Gold Funds, plus the additional 330 bps increase delivered by the Concentrated Portfolios, plus an incremental 130 bps due to Ensemble Methods.

Conclusion

How can the investment industry leverage these insights? Our original white paper described several ways to deliver EAM-based solutions. De novo solutions featuring a “full-Ensemble” can be launched, while, alternatively, existing active portfolios can be readily modified to incorporate “mini-Ensemble” elements.

But whether building new investment solutions or upgrading existing ones, the critical point is that the construction of the Alpha Engine and the construction of the Beta Anchor are two discrete activities with a goal to minimize the Beta Anchor and its dilutive effect on performance.

The beauty of EAM is that the added layer of diversification de-risks the Alpha Engine before pairing it with the Beta Anchor. Therefore, the Beta Anchor’s role is automatically reduced and the natural benefits of the Alpha Engine are delivered directly to the investor.

If EAM portfolios can deliver on their early promise, then Ensemble Active Management may prove to be the disruptive innovation that the active management industry has been searching for and thus could redefine the competitive balance between active and passive management.

For today’s active managers, such a breakthrough cannot come fast enough.

If you liked this post, don’t forget to subscribe to the Enterprising Investor.


All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer.

Image credit: ©Getty Images/CSA-Archive

About the Author(s)
Alexey Panchekha, CFA

Over his nearly three-decade-long career, Alexey Panchekha, CFA, has spent 10 years in academia, where he focused on nonlinear and dynamic strategies; 10 years in the technology industry, where he specialized in program design and development; and eight years in financial services. In the latter arena, he specialized in applying mathematical techniques and technology to risk management and alpha generation. For example, Panchekha was involved in the equity derivative trading technology platform at Goldman Sachs, and led the creation of the multi-asset multi-geographies portfolio risk management system at Bloomberg. He also served as the head of research at Markov Process International, a leader in portfolio attribution and analytics. Most recently, Panchekha co-founded Turing Technology Associates, Inc., with Vadim Fishman. Turing is a technology and intellectual property company that sits at the intersection of mathematics, machine learning, and innovation. Its solutions typically service the financial technology (fintech) industry. Turing primarily focuses on enabling technology that supports the burgeoning Ensemble Active Management (EAM) sector. Panchekha is fluent in multiple computer and web programming languages and software and database programs and is certified in deep learning software. He earned a PhD from Kharkiv Polytechnic University with studies in physics and mathematics as well as an MS in physics. Panchekha is a CFA charterholder.

Matthew M. Bell, CFA

Matthew M. Bell, CFA, is president of Bell Family Interests, a private family office management and consulting firm. He is also an active investor in private companies through direct transactions, and a founding member of the Alamo Angels Network. Prior to founding Bell Family Interests, he was chief investment officer (CIO) and director of family office services for Cross Financial Services Corporation, a Texas-based financial planning and investment management firm. Among his other key positions, were president of Southwest Investment Management, a Texas-based registered investment advisor (RIA) firm, and managing director of The Trust Company, N.A., a federally chartered trust bank headquartered in San Antonio, Texas. Bell also served as the president of the Financial Planning Association of San Antonio and South Texas chapter. He is a graduate of Southern Methodist University with a BBA in finance. He is a CFA charterholder and Certified Financial Planner.

Robert S. Tull, Jr.

Robert S. Tull, Jr., has been a well-recognized expert in the exchange-traded fund (ETF) market since 1993, when he was one of the principals behind the development of WEBS, the precursor to iShares ETFs. Since then, he has consulted with issuers and governments on ETF infrastructure support, become a named inventor on multiple security patents involving exchange-traded products, and played a leading role in the design and development of over 400 exchange-traded products in the United States, Europe, and the Pacific Rim. Recently, Tull was presented the ETF 2018 Nate Most Lifetime Achievement Award. He is one of the founders of ProcureAM and previously was the owner and primary consultant of Robert Tull & Company. Prior to launching Robert Tull & Company, he held senior level roles at such firms as Morgan Stanley; Deutsche Bank, where he was managing director and COO of the Bankers Trust Global Custody, Benefit Payments, and Master Trust business units; and the American Stock Exchange (AMEX), where he was vice president of new product development and executive director of AMEX ETF Services.

3 thoughts on “Ensemble Active Management (EAM): Taming Toxic Tails”

  1. David Fuhrman says:

    How does EAM differ from having several managers, each managing a sleeve of the portfolio, similar to the American Funds approach to active management?

  2. Nice advertisement placement. Contact Ken Fisher; he may have work for you.

  3. Matthew M. Bell CFA says:

    David (Fuhrman), thank you for a question that speaks to the importance of making a distinction between the possible value that can be captured by embracing the manager sleeve approach versus the ensemble active management (EAM) approach.

    The traditional approach of constructing a multi-manager portfolio that is made up of sleeves of individual managers or management teams typically incorporates into each sleeve their full portfolios with their Beta Anchors included. As a result, the Alpha Engines of each sleeve remain diluted by each sleeve’s Beta Anchor. EAM combines active the Alpha Engines only (i.e., the high conviction portion or active share of the portfolios) and collapses the sleeves into one portfolio that is made up almost entirely of their active share Alpha Engine names. EAM combines the managers’ Alpha Engines such that the individual manager biases serve as an additional source of diversification, thereby minimizing (or even eliminating in some cases) the need for a Beta Anchor.

    Both approaches, EAM and the Beta Anchor, achieve similar results from a risk reduction standpoint. But the traditional approach, employing a Beta Anchor, simultaneously dilutes dramatically the active returns that are otherwise available from the Alpha Engines. EAM preserves the active share returns available in the portfolio and even slightly enhances them through consensus-based diversification. As a result, the active returns that are currently being diluted away by Beta Anchors become available to be delivered to the investor. Using different estimates of the active share of average mutual fund portfolios, we estimate that this dilution wastes 200-300 bp annually.

    To be clear, employing the sleeve approach to portfolio construction can often lead to benefits such as reduced risk and diversification by investment style or bias, but this approach pales in comparison to EAM in terms of overall capture of the active share performance available in the undiluted Alpha Engines of the portfolio, while achieving the same or better risk reduction and diversification by style or bias.

Leave a Reply

Your email address will not be published. Required fields are marked *



By continuing to use the site, you agree to the use of cookies. more information

The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.

Close