Bond Analytics for Relative Value Strategies
Relative value trading is a popular investment strategy for firms looking to achieve high returns while minimising risk. This blog explores the challenges of bond analytics and how access to the right analytics can provide opportunities for more comprehensive trading strategies.
27 Jan, 2022

Relative value trading is a popular investment strategy for firms looking to achieve high returns while minimising risk. This blog builds on a previous whitepaper published by Quantifi – Growth of Relative Value Credit Strategies – which explored how an increase in bond issuance (as was the case in 2021), combined with extreme volatility in credit markets, creates attractive opportunities for relative value credit strategies.

Relative value credit strategy depends on the isolation of identical or very similar credit instruments, where one is assessed to be comparatively under- or overvalued. These might be bonds issued by the same borrower but at different points of the yield curve or be bonds issued by different but similar borrowers. For instance, straightforward credit analysis of cash flow and balance sheets of two seemingly similar pharmaceutical firms might reveal that one deserves to be trading at tighter yields than the other, which in turn invites a long/short trade. In essence, the strategy relies on extracting value from dislocation in the pricing of credit risk across markets, instruments and maturities. Although these strategies can differ significantly from each other, they have one thing in common; that is to take advantage of the opportunities requires sophisticated bond analytics. This whitepaper explores the challenges of bond analytics and how access to the right analytics can provide opportunities for more comprehensive trading strategies.

What are the biggest challenges for credit analytics?

With the rapidly evolving financial landscape and radical transformation of fixed income investing, firms have to deal with various analytics challenges. A survey jointly conducted by Quantifi, Celent and 7Chord – as part of a webinar on ‘The Evolution of Credit Trading’ revealed that advanced modelling (29 percent) and consistent relative value (29 percent) rank as the biggest obstacles in credit analytics. Today’s credit investors need reliable data and powerful analytics to help them gain actionable insights for better portfolio outcomes. The ability to anticipate and respond to market and portfolio changes are key motivators for investment managers to maintain a strong risk function. Sophisticated and robust analytics are an important component of this capability.

What do you consider the biggest challenge for credit analytics?

Bond Analytics

The importance of advanced bond analytics for successfully trading relative value credit strategies is well understood, but what are bond analytics?

One can successfully trade fixed and floating bonds based on yield and discount margin, as well as their first (duration) and second (convexity) derivatives. A more comprehensive take on the bonds performance and associated risks requires a more complex measure, namely Z-spread. Z stands for zero volatility and is defined as a flat spread over the discounting curve, which replicates the bond price. Note that while yield calculations are completely model-independent, calculating Z-spread requires the discounting curve to be defined and built first. Since the early 2010s the market standard for discounting was to use OIS curves rather than LIBOR. This, however, presents the challenge of having to model two IR curves for floaters – one for discounting (for example, Fed Fund) and another for projection (for example, USD LIBOR – to be replaced by Secured Overnight Financing Rate (SOFR). To gain consistency with credit markets, one has to imply bond equivalent CDS spread, which is a flat CDS spread of the credit curve used to replicate the bond market price. This requires building IR and credit curves, making this calculation strongly dependent on corresponding modelling assumptions.

Bond Analytics for Callable bonds

The aforementioned approaches to pricing bonds have been used for years. However, faster moving and more complex markets have increased the demand for sophisticated analytics. This is particularly relevant for callable bonds, which are bonds (typically corporate) with an embedded option for the bond issuer to call on specified dates at specified prices (note that bonds can also be puttable when the option exists for the bond buyer). Accurate valuations and sensitivities are key components of relative value credit strategies. This is because many firms issue both regular and callable bonds, and having advanced analytics allows anomalies in market prices to be captured.

Callable bonds create a challenge for yield calculations because the maturity of the bond is not well defined. This requires not only the yield to maturity to be calculated but also yield to first, second, worst and so on. Calculating Z-spread, which in the case of callables is referred to as Options Adjusted Spread (OAS), requires an IR model, ranging from the simple short rate model (Hull–White, Black–Karasinski) to the more comprehensive forward rate model (Brace Gatarek Musiela) to be implemented. Note that OAS spread is usually assumed to be non-stochastic.

In the current environment of high and volatile credit spreads, having only a one-factor IR model is deemed inadequate. This is particularly true for callable floaters and hybrid bonds. The latter has two distinct payment periods – the first is when the bond pays fixed rate and is non-callable, and the second is when the bond pays floating rate and becomes callable. These bonds are referred to as “fixed-to-floaters”. To achieve a more accurate pricing of these bonds, one needs to assume that the credit component is stochastic. The most effective way to do this is by implementing a two-factor IR credit model, where IR and credit components are correlated. Graphs 1 and 2 display how rates and credit volatilities effect prices of callable floater and callable fixed bonds. Both the bonds are par, and option is struck at par. It is clear that rates vols are not effecting the floater bond call, whereas the credit vols are effecting both the fixed and floater calls.  Therefore implementing credit as stochastic is important not only for floaters but fixed bonds as too.

Graph 1: Effect of Rates normal vols (in bps) on a value of a Call option with Strike 100 for par Fixed Bond and par Floater bonds
Graph 2: Effect of Credit normal vols (in bps) on a value of a Call option with Strike 100 for par Fixed Bond and par Floater bonds

Hybrid bonds require further model enhancements, including step-up spread for floating period, and the use of swap spread as a floating rate. Hybrid bonds have very long maturity times, often more than 50 years, with many years as a callable period. This therefore requires a very efficient implementation of a callable model.

This blog was taken from a whitepaper published by Quantifi titled ‘Take Advantage of Relative Value Credit Opportunities with Advanced Bond Analytics’. This excerpt covers the challenges for credit analytics, bond analytics for callable bonds. The full whitepaper covers calibrating credit curve to bonds, the advanced analytics required for convertibles and the LIBOR replacement.

Let's Talk!

Schedule a personalised demo today

Loading...