The New Frontier in Fixed Income: Systematic Investing

Because systematic approaches depend on different performance drivers, their returns will likely be different from—and complementary to—traditional active strategies. The active returns from security selection in systematic strategies are by design largely uncorrelated both with the benchmark and with important risk premia. As a result, these strategies can be effective diversifiers in a fixed-income portfolio.

In addition, systematic approaches can be readily customized, allowing managers to tailor portfolios exactly in line with client preferences while retaining performance potential.

How Do Predictive Factors Work?

Risk factors, such as interest-rate risk (duration) or spread duration (the sensitivity of a bond’s price to changes in its credit spread), identify the ways that market drivers can influence security prices. Systematic strategies aim to discover factors that have predictive power to repeatably find securities with the best risk-adjusted return potential—that is, predictive factors. These can be market value–based (for instance, value and momentum) or fundamental, company-specific factors (such as quality). Predictive factor insights are used to systematically analyze large volumes of historical market data and single out securities with the right characteristics to have an above-average probability of outperforming the market (Display).

Managing Factor Strategies in Practice

Leading systematic managers have deep research databases and in-house cutting-edge quant platforms, enabling them to identify and access hundreds of proprietary factors. Not all will be implemented continuously in a systematic portfolio, but the managers can rotate them according to market conditions as investment regimes change and the efficacy of the factors varies with them.

Predictive factor-based approaches originated in equity markets, where benchmarks are relatively straightforward to construct, and pricing is largely transparent. Factor-based approaches arrived more recently in fixed-income markets, which are larger, more complex and fragmented across disparate trading pools. All these features make liquidity and pricing harder to discover in bond markets.

For these reasons, advanced technology and analytics are vital to making systematic approaches work in fixed-income markets. And although academic research supports the case for predictive factors in fixed-income investing, it takes rigorous testing and practical implementation skills to create successful portfolios.

Creating a Systematic Portfolio: Combining Predictive Factors

With a systematic approach, each bond in the benchmark is scored on a range of predictive factors. This results in an array of scores for each security. For instance, a bond might have a high score on value but a low score for momentum. A factor combination model then rolls up the different factor scores to produce a single composite total factor score for each security.

The model adopts two criteria to create a portfolio using the factor scores: predictive efficacy and correlation with other factors. It weights them using an algorithm determined by a machine learning technique. This ranks the total factor score for each bond subject to other optimization and risk constraints, principally: bond, issuer, sector, ESG, duration, spread, liquidity and transaction cost limits. In this way, the model seeks superior risk-adjusted returns by balancing predictive efficacy with rigorous risk controls.

Three Critical Success Factors

In this relatively new field, the performance of some providers’ systematic fixed-income products has proved disappointing, with live returns failing to live up to back-tested results. We think there are three frequent failings: relying on static factors, using unreliable data, and an inability to source liquidity and to execute ideas effectively. These illustrate the importance of three “pillars” for effective systematic strategies:

A Dynamic Factor Approach. Market conditions are always changing, and factor efficacy varies from market to market and over time. For instance, carry (yield) may be a strong factor in investment-grade bond markets but not in high yield, where default risk is a more important performance driver. Consequently, it’s crucial to evaluate factors continuously and manage them dynamically.

Abundance of Data. Reliable data are the indispensable building blocks of effective systematic strategies. Systematic investors need vast quantities of data that are clean (free of anomalies and inconsistencies), extensive and have a very long history. Compiling those data is a laborious, research-intensive task. It involves amassing data that extend across a wide range of bond metrics, include point-in-time analytics for companies’ financials across multiple fields and cover many fixed-income classes globally.

Liquidity Considerations. Firms that can’t effectively assess a bond’s liquidity won’t be able to implement their investment ideas. To keep up in a marketplace that will digest and react to every new bit of information faster and faster, successful fixed-income managers need to use technology that pulls all external fixed-income trading platforms together in one place. Finding adequate liquidity to execute desired trades is a precondition for managing a portfolio’s factor weights. And finding enough liquidity at attractive prices is paramount, because systematic strategies only execute trades if they pass tests for transaction-cost effectiveness.

Questions to Ask Your Systematic Manager

Does your process manage factors dynamically, and how many factors does it use?

We find that systematic strategies using a dynamically weighted and wide range of factors are positioned to deliver better outcomes than those that rely on fewer factors and that use static factor weights.

Have you embedded liquidity analysis in your investment process?

Compared with equity trading, fixed income is much more problematic: more manual, less transparent and less liquid. That’s a problem, especially for equity managers attempting to transition to bond investing.

Consequently, it’s important to assess liquidity at the start. Advanced systematic approaches incorporate liquidity information to help source the exact bonds their model selects to buy. Overly basic approaches may send a list of bonds that meet certain criteria to their trading desk and buy those they can source—even if they’re not their model’s top choices. And if there aren’t enough sourceable bonds to buy, trading may be delayed as the process starts over again.

How extensive are your data?

Abundant, clean data compiled rigorously and covering many years’ history are the foundation of robust systematic investing. Beware managers that have short data histories and/or have bought in lower-quality data sets.

How do you integrate AI into your process?

Managers can add value with machine learning techniques at many levels of a systematic process.

AI enhancements range from time-saving and efficiency gains (identifying price patterns to impute missing data quickly and reliably) to qualitative advances (improving analytics across multiple valuation factors to help find new signals and make existing signals more effective). Managers that don’t harness the power of AI will quickly be left behind.

Is your investment approach exclusively quantitative?

We believe that integrating quantitative experts within a broader fixed-income team that includes fundamental active bond professionals such as portfolio managers and traders can bring big benefits. This broader experience can improve execution, thus lowering transaction costs, and provide a wider perspective, practical insights and sanity checks to help evaluate factors and fine-tune models. The team can also highlight risks and opportunities that might not have been previously back-tested, for instance, the benefits of incorporating the new issue market into a systematic process.

Does your process create a black box?

Systematic fixed income is an active approach that depends on discovery, selection and monitoring of predictive factors and that needs human input. While a model determines the factor weights and is instrumental in back-testing the data, human involvement is critical in testing the factors and deciding to add new ones and remove others.

Your manager should demonstrate not only an understanding of factors but also the ability to identify new factors, to evaluate factor performance and to modify the portfolio’s factor mix as conditions change.

How would you describe your results?

Investors should expect robust systematic approaches to exhibit lower credit risk than traditional active products because their focus is more on security selection, and they control sector and credit risks more rigorously. For the same reasons, we expect that strong systematic managers will also exhibit better upside and lower downside capture, with high information ratios.

An Idea Whose Time Has Come

We believe systematic fixed-income investing is an idea whose time has come. It provides an active way to achieve attractive, repeatable and uncorrelated risk-adjusted returns through:

• Bottom-up selection and structuring of many independent holdings
• A high level of risk control that makes systematic portfolios relatively less vulnerable to big drawdowns resulting from adverse interest-rate, credit and other single-factor events

This dual emphasis on outperforming through individual security selection and rigorous risk control makes systematic approaches complementary to most traditional active fixed-income strategies, and potentially powerful portfolio diversifiers.

Investors today face a world where conditions are rapidly changing, with increased availability of data and the advent of AI shifting familiar paradigms. We believe a leading-edge systematic approach that uses these new developments can help provide an objective, evidence-based path to more consistent risk-adjusted fixed-income returns.

For further information, read our white paper, Systematic Fixed-Income Investing Comes of Age.

The views expressed herein do not constitute research, investment advice or trade recommendations and do not necessarily represent the views of all AB portfolio-management teams. Views are subject to revision over time.

A message from Advisor Perspectives and VettaFi: To learn more about this and other topics, check out some of our videos.

© AllianceBernstein

Read more commentaries by AllianceBernstein