Portfolio diversification is a fundamental concept in investment management, aiming to optimize risk-adjusted returns by combining assets with low correlations. Unlike the equity market with a long tradition of building such portfolios based on well-founded analytic methods, there are no such systematic approaches in building corporate bond portfolios. Rather, portfolio managers have relied on traditional judgement approaches based on constraints including the percent allocated to industry sectors, issuers, and ratings. In this article, we will consider several alternative analytic approaches for building corporate bond portfolios and compare them over a long period of time. Our focus in this article will on the US High Yield corporate bond market. In future work coming soon, we will extend our empirical analysis to the US Investment Grade corporate bond universe as well as to the European corporate bond market.
The alternative approaches to corporate bond portfolio construction we will be analyzing are:
· Equally-Weighted
· Maximum Diversification Ratio (DR)
· Rolling Mean-Variance-Entropy (RMVE)
In this article, we first describe in detail these three portfolio construction techniques. We next discuss the types and variables of corporate bond data collected. Then we will describe how these diversification-based investment strategies are implemented with real corporate bond data spanning the full US High Yield corporate bond universe over an extended time period from 1997 to 2020. We will then show their performance in terms of cumulative P&L, returns, volatilities, and Sharpe ratios over this time period, and discuss the most performing investment strategy among them. Finally, we will compare their performance versus a US corporate bond Benchmark index.
The following is a time series of portfolio weight heatmap (time on the x axis, and sector index on the y axis):
We find a clear advantage for the mean-variance entropy portfolio1:
(see link at the top of the post for complete article.)
In conclusion, we analyzed three different portfolio constructions for High-Yield corporate bonds using systematic criteria, namely: equal-risk, maximum diversification, mean-variance entropy. All three approaches beat the BoAML High-Yield benchmark. We also found that, overall, the mean-variance entropy portfolio has superior performance relative to the other methods as well as in absolute terms.
Quantitatively Yours,
mu and sigma are respectively the mean and standard deviation of the monthly returns; and SR is the annualized Sharpe Ratio with zero rate = sqrt(12) * mu / sigma