Building a Portfolio: A Closer Look at Our Process

One of the most important elements of our systematic strategies is our collaboration with Barclays Quantitative Portfolio Strategy (QPS), an organization that has long been recognized as one of the leaders in credit quantitative research. Each day, Barclays QPS provides individual signal scores for three alpha factorsfor each bond in a specified index. Each signal is scored on a scale of 1-10 for each factor and we used a combined score of the three signals to rank index constituents. Our process seeks to maintain exposure to high scoring bonds.

In this post, we discuss our goals for the construction of portfolios and the choices behind the signals used.

Our Goals

We designed our portfolio construction process to aim for the following goals:

• Generate alpha using systematic, data-based signals. We believe that signals derived from quantitative data can reveal insights into future price performance—including trends that traditional, fundamental analysis may miss. We are using these signals, which pull information from both credit and equity data, to identify bonds that can potentially generate excess return versus benchmarks.
• Eliminate additional beta risks. We aim to limit top-down deviations relative to the benchmark, such as duration, quality and sector, to focus on the output of these signals, which is reflective of individual bond and issuer scores.
• Incorporate liquidity screening. With an eye toward implementation success, we screen out bonds with below-median liquidity.

At bottom, we want to be consistent with the “client-first” mindset that drives our team’s decision making. We need to make sure that the systematic portfolios are generating alpha, are in line with client objectives and are measured against an appropriate benchmark.

Putting the Right Signals to Work

The first systematic strategies coming out of our collaboration with QPS are focused on investment grade corporate credit where strategies are based on signals related to three factors: Value, Momentum and Sentiment.

1. Value, which is tightly connected with income, is a critical factor in credit selection, as it seeks to identify bonds that allow investors to realize income or return relative to peers. The signal used in our process is called Excess Spread over Peers (ESP). This can also be thought of as “quality-adjusted spread.” Each issue is evaluated in the context of peer groups. Peer groups are based on sector/industry, quality, rating and maturity.
   
   ESP is a relative value signal designed to identify bond mispricings after accounting for fundamentals. Data shows that credit quality beta is the key to outperformance over time, so this signal looks to maximize the spread for a within the context of issuer quality. Historically, portfolio comprising high ESP-scoring bonds have consistently had higher average excess returns and information ratios than those with low ESP-scoring bonds.
2. Momentum. Second, we include a momentum signal, which is Equity Momentum in Credit (EMC). This is derived based on trend patterns of an issuer’s equity share price over 1-, 3- and 6-month periods.Stock prices have shown to be strong indicators of credit momentum. Data shows that trending stocks have stronger EMC signals, and it’s important to look at whether the stock is simply trending, or actually volatile.
3. Sentiment. Third, we include a sentiment signal, Equity Short Interest (ESI). This is another cross-factor signal which is based on short equity positions outstanding, controlling for the limits on borrowing imposed by the securities lending market. ESI has some asymmetry to it; the majority of issuers will score well on this signal, but for the few that score poorly, the issuer is much more likely to experience low subsequent equity and credit spread returnsand should be avoided. Barclays QPS has documented that the predictive power of ESI extends to a firm’s corporate bond returns.

For details on our backtesting data, please contact the team.

Risk Considerations

To implement our systematic portfolios, we rank each bond in an Index by combined signal score and use this in an optimized portfolio construction process. An important step is that we incorporate risk controls in an effort to remove the impacts of certain outside influences on bond performance. On a daily basis, we review the duration-times-spread (DTS), the duration, the sector weightings, the ratings, and the issuer concentration of our systematic portfolio versus the Index, and make sure that the portfolio remains within specified ranges on those parameters. For example, we aim to keep duration +/- 0.25 years of the Index, and sector exposure within +/- 3% of the benchmark sector weights.

Furthermore, we want to minimize costs, and transaction costs associated with turnover can be a significant contributor to implementation costs. Our research indicates that over time the benefits associated with portfolio turnover may be offset by excessive transaction costs as turnover rises above 12-15% per month. So all else equal, we want securities with less turnover.

We are cognizant of transaction costs and the negative impacts they can have on portfolio outcomes. Our analysis shows that over time, portfolios that average approximately 10% monthly turnover strike the right balance between actively seeking opportunities while avoiding excess transaction costs. Therefore, our optimizations cap turnover at 10%.

Implementation

By incorporating risk and liquidity constraints, we aim to replicate the optimized portfolio with the same risk parameters, while minimizing costs and increasing incremental alpha (Figure 1).

State Street Global Advisors Capabilities

Through our collaboration with Barclays QPS, we can use robust data to designate which securities are optimal for the systematic portfolios. However, transparency and pricing needs to be considered as well. Fixed income markets are broad and fragmented, and even though the asset class has made great strides in trading efficiencies, illiquidity remains a concern for certain issues.

Our capabilities as an index manager allow us to efficiently build portfolios that tightly control signal exposure while optimizing liquidity and minimizing trading costs. We combine the Barclays QPS data excellence with our practical portfolio sampling and implementation experience.