Credit quarterly newsletter: How data analysis is vastly improving our Credit team’s investment process
The stock markets provide a vast array of opportunities for investors, but the credit markets are even more complex to navigate. That’s because any issuing firm could have multiple bonds outstanding with different maturities, bond formats or different places in the capital structure. This quarter we talk about some of the data-analysis tools we’ve developed to help us pinpoint the best opportunities for our clients among all the opportunities available in the market.
Our Credit team uses a disciplined investment process combining top-down and bottom-up analysis. Our top-down analysis looks into the trends and themes that are driving the markets and sectors. Based on this analysis, we determine the level of risk to take in the portfolio and how to allocate it across sectors. Individual security selection is the result of our bottom-up analysis, which involves looking at company fundamentals and taking relative-value positions. We make extensive use of data analytic tools to help us with relative value analysis. Let’s look at how this works.
A wide universe to cover
The eight members of our Euro Credit team all cover different sectors, but they all typically analyse over 60 companies and between 300–600 bonds, including both investment-grade and high-yield issues. This clearly has scope to be a hugely time-consuming process.
We start off by ranking the companies in each sector according to their relative quality (e.g. leverage, management, ESG etc.), and from there we analyse how attractive they are by looking at their bonds’ spreads. While company fundamentals are in general evolving relatively slow, credit spreads change all the time. It’s therefore vital for us to constantly scan the universe to ensure that we’re still invested in the most attractive bonds and not missing out on more compelling opportunities.
Previously we used a combination of tools to follow the bond universe. Each portfolio manager had their own spreadsheets and formats, an exercise that was largely manual. What’s more, the team used Bloomberg to look at scatter plots and make relative value analyses. Finally, the information on how the portfolio was positioned was only available in our front office order management system, thinkFolio. Using different tools and sources of information manually was clearly not very efficient.
To gain efficiency, make our infrastructure more robust and improve idea generation, we’ve developed an in-house system called SCARLETT. This tool combines portfolio, benchmark and market data from different sources and stores spread levels and other characteristics for all the individual bonds in our universe. We complement this with other data like credit default swap spreads to be able to monitor different sources of market information. SCARLETT shows our portfolios’ positioning and calculates spread and price developments over the past day, week, month and quarter. It also aggregates bonds by characteristics such as their rating, sector, and maturity buckets, making it possible to compare the relative attractiveness of an issuer within its sector or rating. Results are displayed in interactive dashboards that the portfolio managers can filter in many ways to make any combination of variables possible.
All this results in vastly improved efficiency for the portfolio managers, significantly increases the number of trade opportunities that we identify, and improves the performance potential of our credit strategies.
Let’s look at some examples of the kind of functionality that SCARLETT provides.
Easy comparison of bonds
The screenshot above shows one of the SCARLETT dashboards. On the left we can see a table of bonds with their main characteristics. It also shows the changes in their spreads over various timeframes. In the scatter chart at the top-right we plot each bond’s spread against its maturity, enabling us to quickly identify how bonds are trading against each other. The bonds that we own are shown as filled circles and green bonds are represented by squares. If we click on a bond, a graph of that bond’s historical spread relative to the benchmark appears, enabling the team to quickly see how a bond has performed. The dashboard is interactive; the table can be sorted by column, and on the right-hand side there is a set of filters to browse through the data.
“In just one dashboard we can now see information that was previously spread across different systems. This makes finding relative value opportunities far easier and less time-consuming.”Quirijn Landman, Senior Portfolio Manager Credits
Tracking new issues
It’s common for companies to issue new bonds in the primary market. Usually these new issues are announced in the morning, and the team needs to decide if it wants to participate by the end of that morning. This means there is limited time to assess the attractiveness of the new issues. Again, SCARLETT makes it very efficient to do so.Let’s say a firm with an A credit rating issues a new bond with a ten-year maturity. During the morning the issue is announced we use SCARLETT to check if it represents good value relative to that company’s outstanding bonds, and also relative to other bonds in the same sector with similar characteristics. Additionally, we are now able to easily compare the bond with others in different sectors – in other words, bonds not covered by the portfolio manager looking at the new issue. By applying filters in the dashboard we can look at, for example, all A-rated bonds with a maturity of around 10 years in our universe, providing much more insight into the attractiveness of the new issue in a broader context. Previously we would have had to ask all other team members to provide this input within a very tight timeframe. This saves more time to dig into the company’s fundamentals, and leads to a better informed decision.
Customised emails highlighting opportunities of interest
Every morning the SCARLETT system provides each portfolio manager with a customised email detailing the biggest spread movements in the bonds in the sectors they cover. The system uses a machine-learning algorithm called clustering that groups together securities with certain similarities. The system shows the most significant moves for each cluster, providing a useful summary of the most interesting bonds to look at.
Tracking spread developments over time
As we’re storing data on a daily basis we can easily show how spreads change over time. This is also possible in Bloomberg, and in the past we might have seen that a security we own had performed well relative to another, prompting us to switch into the second security because the spread differential had widened. We perform a lot of these kinds of relative value trades.
But in Bloomberg we couldn’t plot the change in spread of a particular bond against the change in spread of the overall benchmark or a particular sector. SCARLETT enables us to do this.
In the chart above we see a bond issued by Aeroport de Paris (ADP). The dark blue line shows the spread history of the ADP 2028 maturity bond over time and the orange line shows the spread history of the infrastructure sector within the iBoxx € Corporate benchmark. We can see that the spread difference between the two was around -70 basis points (bps) before the Covid-19 pandemic, indicating that ADP was trading at a much lower spread than the average spread for a bond in the infrastructure sector. But as countries locked down their economies in response to the pandemic, the spreads of ADP bonds widened. At the beginning of April the spread difference had increased to +60 bps. Around this time we started adding ADP bonds to our portfolio as we perceived it to be good value relative to the broad infrastructure sector. Since then the spread difference has fallen again, showing that the bond has outperformed the overall sector.
And we can also play around with the data: ADP has an A credit rating, so we can compare its performance over time with that of other A-rated names in the index. Previously, it wasn’t possible to do this in Bloomberg – it had to be done manually in Excel, which was very cumbersome.
Our system has really improved our understanding of what’s going in the market, helping us to find opportunities in a far more efficient manner.
As we stated above, each portfolio manager analyses between 300-600 bonds on average. Looking at relative value within this set of bonds leads to more than 100,000 trading combinations for each portfolio manager to look at within their sectors. This figure does not even take into account bond pairs across sectors covered by different portfolio managers. Taking those opportunities into account increases the number of possible combinations to more than 4 million. It is of course impossible to look at all these combinations, meaning we could miss out on interesting opportunities.
Example of output Turbo Matrix dashboard.
To solve this, we have developed a system that analyses spread developments in our bond universe, and looks at individual bond pairs. This system is called the Turbo Matrix and it analyses over 4 million pairs every day, helping us to find the most interesting trading opportunities. It’s completely customisable, so we can choose to filter bonds by characteristics including their sector, rating, maturity, seniority and whether we hold them. We are now able to screen the whole universe every day, ensuring that we see all compelling trade opportunities while at the same time making relative value screening much more efficient.
We look at alternative data sources to provide us with additional and more real-time insights. For example, we follow airline and airport companies. These firms’ businesses deteriorated sharply as a result of the recent lockdowns. As a team we would like to get an indication of how the results of these companies are developing on a regular basis. We could wait for the companies to disclose their figures, but this could mean there is a considerable time lag, as they might only publish their financials or details of their number of flights on a monthly or quarterly basis. It would be very useful for us to find out these trends as soon as possible. So we scrape flight data for airports around the world to track the number of flights over time.
The chart below shows the percentage of flights per airport this year compared with the number of flights in January. As expected, it shows a sharp drop in the number of flights during the coronavirus outbreak. More interestingly, we can now track almost real-time which airports are seeing their number of flights increase. While these figures do not completely correlate with the company’s fundamentals, there will certainly be a strong link with the financial results. This gives the team a better feeling of how fundamentals are developing, and potentially how to position the portfolio in response.
At Kempen we’re continuing to invest in data analytics as we think it will provide additional insight and improve efficiency of our investment processes. For example, we’ve recently subscribed to a service with fundamental data for all the companies in our universe. We plan to integrate this data in SCARLETT to further improve our bottom-up analysis. We also look to include more alternative data sources, which could provide us with additional and more real-time insights. All these initiatives should help in further improving our investment process in the future.
Kempen (Lux) Euro Credit Fund (the “Sub-Fund”) is a sub-fund of Kempen International Funds SICAV (the “Fund”), domiciled in Luxembourg. This Fund is authorised in Luxembourg and is regulated by the Commission de Surveillance du Secteur Financier. Kempen Capital Management N.V. (KCM) is the management company of the Fund. KCM is authorised as management company and regulated by the Dutch Authority for the Financial Markets (AFM).
Paying agent and representative in Switzerland is RBC Investor Services Bank S.A., Esch-sur-Alzette, Zurich Branch, Bleicherweg 7, CH-8027 Zurich. The Sub-Fund is registered with the Dutch Authority for the Financial Markets (AFM) under the license of the Fund.
The information in this document provides insufficient information for an investment decision. Please read the Key Investor Document (available in Dutch, English and several other languages, see website) and the prospectus (available in English). These documents as well as annual report, semi-annual report and the articles of incorporation of the Fund are available free of charge at the registered office of the Fund located at 6H, route de Trèves, L-2633 Senningerberg, Luxembourg, at the offices of the representative in Switzerland and on the website of KCM (www.kempen.com/en/asset-management). The information on the website is (partly) available in Dutch and English.
The Sub-Fund is registered for offering in a limited number of countries. The countries where the Sub-Fund is registered can be found on the website. The value of your investment may fluctuate. Past performance provides no guarantee for the future. The performance shown does not take account of any commissions and costs charged when subscribing to and redeeming units.
This document is prepared by the fund managers of the (Lux) Euro Credit Fund, managed by Kempen Capital Management N.V. (‘KCM’). The (Lux) Euro Credit Fund might currently hold bonds in the subject companies. The views expressed in this document may be subject to change at any given time, without prior notice. KCM has no obligation to update the contents of this document. As asset manager KCM may have investments, generally for the benefit of third parties, in financial instruments mentioned in this document and it may at any time decide to execute buy or sell transactions in these financial instruments.
This document is for information purposes only and provides insufficient information for an investment decision. This document does not contain investment advice, no investment recommendation, no research, or an invitation to buy or sell any financial instruments, and should not be interpreted as such.
This document is based on information that we consider reliable, but we do not represent that it is accurate or complete, and it should not be relied on as such.
The views expressed herein are our current views as of the date appearing on this document. This document has been produced independently of the company and the views contained herein are entirely those of KCM.