For Vuk Magdelinic, using artificial intelligence in corporate bond trading is more than a promise — it’s a reality.
“We can automate 30 percent of a trader’s entire trading workflow,” says the CEO of Overbond, a fixed income AI analytics company. “So, if I’m a trader and I trade 300 transactions a day, 100 out of 300 I never see again. They literally never come across my screen and they are executed with speed and precision. They reflect my book style and bias, just like I was executing them. Imagine 30 percent of everything I do just goes away, but it still earns money to my book equally.”
It’s hard not to share his enthusiasm. A decade ago, if you asked corporate bond traders about the possibility of trading corporate bonds electronically, they almost all said it could never be done. That it would just be too complicated. But it’s happening. And it’s growing. Even more surprising is that advances in technology are facilitating applications of AI that allow automated trading.
There are, however, unique challenges to automating the bond market, such as the diversity of fixed-income issues. A single company might have several outstanding bond issues, each with unique coupons, times remaining to maturity or other features — and liquidity could change as they age. Another challenge is the illiquidity of bond issues, since many are small in size and trade rarely, if at all.
Because of these structural issues, it remains an over-the-counter (OTC) market, where participants trade directly between participants. Without the centralization and transparency of an exchange, an OTC market is difficult to automate — which is why traders long believed it would remain so.
Owing to the lack of a centralized data source, the corporate bond market is notoriously opaque. Buy-side and sell-side participants and issuers can all benefit from better real-time data. The ability to generate this alone is valuable. But it’s also the first hurdle that must be overcome for any automated trading system or meaningful analytics.
With no universal data provider or transparent data such as an exchange, creating a useable data set falls to the firm. But aggregating, cleaning, normalizing and de-duplicating data is extremely difficult and resource-intensive.
Overbond tackles this problem through client-side data aggregation, which involves aggregating data feeds from various providers such as composite data feeds, transcribed data from chats and voice, records of past trades in order management systems, and post-trade regulatory feeds. Up until a few years ago, it wasn’t possible to process, map, or normalize all these different data feeds in real-time. Today, AI analytics makes it possible.
Overbond’s corporate bond information (COBI) pricing AI compiles this data, optimizes it across balance sheet data gaps, and applies historical benchmarking and curve fitting. COBI pricing AI can build curves for more than 10,000 issuers in various real-time liquidity scenarios.
Advancements in related technologies also allow Overbond’s AI to achieve the speed needed for automated pricing.
“Real-time processing of live streaming APIs and AI-enhanced bond price modeling that demands historical benchmarking and analysis was not possible without the application of cutting-edge, serverless-cloud technology that AWS recently launched,” Magdelinic says.
“Overbond’s real-time AI pricing engine leverages AWS development infrastructure to refresh fixed-income security prices in under three seconds to provide the best executable prices for traders.”
Its AI engine also uses parallelization, which works by breaking up large problems into smaller problems that are then solved at the same time, rather than serially. This allows the pricing program to run through multiple data sets and optimizations with greater speed.
Overbond’s request-for-quote (RFQ) system with auto-pricing capabilities is built on top of the COBI pricing engine. The model can then be “trained” to customize pricing based on the desk’s trading style and client risk preferences.
In addition to pricing capabilities and automated trading, AI offers improved analytics for traders. For example, Overbond provides a live intraday liquidity score on individual bonds that can be used for pre-trade decision-making or to assess hedging options for credit market-makers. It can also quantify liquidity and be used to assess secondary market tradability on any ISIN/lCUSIP, at any time. It does this by considering seven or more variables, including recent trade clustering, which can help determine whether it’s a buyer’s or seller’s market.
Beyond their doubts as to whether automated trading of bonds is even possible, traders are also understandably concerned about AI taking their jobs. But instead of replacing them, AI frees up their time from more mundane trades and allows them to focus on more complex or profitable trades. And analytics gives them a competitive edge in spotting opportunities in the markets and managing risk.
From generating useable data to helping execute trades, there is tremendous potential for AI to be a powerful tool that can help corporate bond traders boost profit and manage risk.
Overbond specializes in custom AI analytics development for clients implementing trade automation workflows, risk management, portfolio modeling, and quantitative finance applications. Overbond supports financial institutions in the AI model development, implementation, and validation stages as well as ongoing maintenance.
Vuk Magdelinic | CEO
+1 (416) 559-7101
Adam Anozy | Sales Associate
+1 (647) 973-4391