Disentangling the Bond Universe with Katana Lens
How would the world be if we were able to make well-informed decisions, if we could evaluate all possibilities and have a universal view before taking the next step? What if our pensions, investments and future were based on this universal thought process?
Androniki Menelaou — Senior Data Scientist at ING WB Advanced Analytics and Product Owner of Katana Lens attempted to answer these questions during her thoughtful presentation at Sibos 2019 which took place for the first time in London, as the capital of a booming tech and financial ecosystem.
Sibos is a renowned conference that connects business thought leaders and decision makers to discuss on the latest innovations and future of finance. This year’s theme, “thriving in a hyper-connected world”, explored the opportunities and challenges that arise from a data-driven era. An era where human and machine competences are leveraged and need to co-exist to bring out the best value propositions for organisations.
In this regard, Androniki’s presentation focused on the interplay between technology and humanity to introduce our anchor product Katana Lens. How can artificial intelligence enhance human’s decision making?
“Humans need to make 35K decisions consciously every day. On the one hand, ING is connected to 38 million people, our retail clients, and many more indirectly through our wholesale clients. On the other hand, ING is expanding its capabilities in technology and big data analytics. We are therefore in the centre of bringing the human need for making decisions and building solutions to enhance and inform their decisions using technology”
In ING, we want to rethink and reshape the decision making processes in the bond market, not only as market makers, but also with our clients from the buy side. We did so by co-creating Katana Lens with the second largest pension fund in the Netherlands, PGGM, and by validating it with more asset managers. Our mission is to augment the decision making process with data analytics and offer one-click executable pair trades. Androniki explained:
“Katana Lens taps into the decision making process of investors that need to decide what bonds to buy and what bonds to sell. Investors will do fundamental research on the company, follow the market trends through the news and perform quantitative analysis on past trends. Given those three elements together, they will make a decision. However, the last part is only explored in 1/1000 because of internal tooling, technology and skillset in the investor companies. Katana Lens breaks the limitations by bringing together both technology and data science to explore 100% of the possibilities”
The data science model behind Katana Lens is based on a statistical anomaly called mean-reversion. Stéphane Collot, one of the data scientists that worked on the model since its conception explained:
“The algorithm is looking for relative value in pair trading. What distinguishes Katana Lens is that it can explore the entire bond universe systematically. With 2000 bonds we have about 2 million possible pairs, and have come up with the best trading ideas”
Katana Lens was developed within ING WB Advanced Analytics from a multidisciplinary squad of data scientists, data engineers and UX designers. The project is based on a human-centered design approach from the very beginning, continuously involving users to each development stage in order to bring out the best version to their needs.
ING WBAA provides the talent and culture to explore and build new propositions and from there, reach the best version of each project.
Learn more by watching the video below: