By Matt Murphy, T. Rowe Price & Venky Vemparala, FlexTrade
As fixed income trading desks cope with multiple trading protocols, venues, and vast data sets of pre-trade liquidity, there is a deeper conversation taking shape around data integration, automation, and interoperability. In this Q&A, Matt Murphy, Fixed Income Electronic Trading and Market Structure Specialist at T. Rowe Price, and Venky Vemparala, Global Product Manager – Fixed Income at FlexTrade Systems discuss the current trends in electronic trading, liquidity aggregation, and adoption of execution management systems (EMSs), as well as the potential for artificial intelligence (AI) to make data-based recommendations.
What are the hot topics being discussed on fixed income trading desks, and how is this different than prior years?
MM: Data, automation, AI, connectivity, liquidity, and interoperability are resonating with the buy side and across the industry. While many of these topics have been discussed over the years, what’s changed is that firms are executing and putting many of these into practice.
VV: At June’s Fixed Income Leaders Summit (FILS) held in Nashville, the discourse was deeper and more concrete than last year with more conversation around how data is used in automation and how automation is used in workflows. Third-party vendors are looking to supplement the EMS, exploring areas like back-office connectivity or post-trade analytics. I take that as a sign of emerging maturity in the industry. An example of this is the ongoing integration that we kicked off between FlexTrade and Intercontinental Exchange (ICE) to bring pre-and post-trade calculations into FlexFI.
The buy side has been talking about electronic trading and aggregating liquidity for a long time, particularly in credit markets. How much progress has been made and what challenges remain?
MM: Over the past five years, there’s been a major uptick in electronic trading. This is apparent in most fixed income markets, especially in credit markets. There has been an increase in trades being executed electronically through various channels. Based on data from Coalition Greenwich, an estimated 40% of US investment-grade bonds and 29% of US high-yield bonds are traded electronically.
There are many different factors that have led to increases in electronic trading including growth in exchange traded funds (ETFs), separately managed accounts (SMAs), and portfolio trading. Increased electronic trading is happening in other less liquid asset classes like municipals and emerging markets.
Another factor is the increased availability of data through APIs (Application Programming Interfaces). A few years ago, the buy-side was not utilising APIs or leveraging a third-party EMS like FlexTrade. However, in recent years there has been more acceptance among dealers, trading venues and other providers to offer their data via an API connection either directly or through an EMS. The availability of data via APIs has made it easier to access liquidity and contributed to the growth of electronic trading.
How is the buy-side leveraging data through the EMS?
MM: Currently the EMS is being used to aggregate market data and liquidity from various sources including trading venues, dealers, and data providers. One of the many benefits to aggregating this data in a central place is to leverage that data to help make data-driven decisions. This data can be used by the trading desk but is also valuable throughout the investment process.
VV: In the beginning, the conversation around data started with basic availability of APIs to connect to counterparties to distribute liquidity in various formats. Today, many platforms and counterparties contribute liquidity directly into an EMS. We work with them to normalise the feed messages into a standard protocol, and we also look for missing or duplicate data fields.
While more work is underway in this area, we can provide clients with a “unique liquidity dashboard” rather than just a “consolidated liquidity dashboard.” In the next stage, I see trading desks gaining insights from such “unique liquidity” to help construct trades or using it in post-trade TCA calculations.
The buy side has access to multiple trading protocols and venues, and some firms want to aggregate direct dealer feeds into a single blotter vis the EMS. How is this going to evolve?
MM: One of the values of an EMS is that it gives our traders access to multiple venues and protocols in a centralised place. Trading venues have built various protocols and trading tools that live inside the venue’s user interface (UI). The EMS can help us access venue tools in a more efficient way, but we realise there will still be cases where we need to use the venue UI. One of our goals is for the integration between the EMS and trading venues to be seamless. One of the ways we are looking to solve this is through partnership and leveraging interoperability.
How is the buy-side selecting the best protocol for a given trade and navigating through that workflow?
MM: This is where data is going to play a critical role both now and in the future. We want to be able to leverage the data that is available to us to help determine what is the best path to execute a trade. One key element is having the data that is in an easily digestible format so it can be incorporated into a trader’s workflow.
VV: There is ongoing layering of automation on top of existing protocols at the major venues. However, protocol selection is a multi-dimensional problem. Pre-trade data and trading history are very much required as part of that selection process, making the EMS a natural place for such automation.
With the complexity of multiple applications running on the fixed income desktop, another topic that’s attracting attention is interoperability. What does it mean for the buy-side trader?
MM: Interoperability is about efficiency and optimisation on the buy side – making workflows seamless and ensuring that applications work together.
Most traders have multiple screens and various applications they utilise daily. When launching a desktop in the morning, traders want their applications and desktop layout to start in the exact same place every day. That’s a simple use case. Though we’re scratching the surface with interoperability, there are many use cases for the trader to increase efficiency and optimise workflow.
Do you have concerns about compatibility if fixed income applications are running in different interop containers or platforms?
MM: The key is the standard language, FDC3, to ensure that all applications speak the same language. This is like today in fixed income where we’re very reliant on the FIX protocol, so having that standard (language) for interop is critical as we start to integrate various applications while new providers continue to enter the fixed income space. We need to connect with multiple applications and different data sources in an efficient manner without adding excessive cost or burdens to either our internal technology team or third-party vendors.
VV: I agree that interop is an efficient way for the EMS to share data and interact with other applications. For interop to be effective in fixed income, we also need to go beyond copy/paste of a bond’s ISIN (International Securities Identification Number) symbol and move into higher-level tasks such as sharing analytics with another application. Another use case is around interacting with chat systems via an interop standard. This will allow a trader to negotiate a bond trade on a chat system that uses interop to book the trade into the EMS.
In a 2023 EMS survey, Coalition Greenwich found that 50% of buy-side firms have adopted or partially adopted artificial intelligence or machine learning (AI/ML) advanced analysis. Does that surprise you?
MM: This doesn’t really surprise me since everyone is talking about AI and the different potential applications. I think most are trying to figure out how to leverage AI and expect this to be a major topic of discussion in the industry over the coming years.
Now that ChatGPT has taken the world by storm, what types of AI-based applications do you expect to emerge?
VV: I am watching several types of AI starting to emerge for bond traders. One type of AI – termed “UX AI” – involves ChatGPT on the EMS dashboard. While we have a core set of APIs to pull in information, with a ChatGPT type of user interface, the trader could generate ad hoc queries in plain English, which are then translated into a combination of API calls We recently launched a prototype using ChatGPT called FlexA, (pronounced Flex-aah, h silent), which allows the user to control the interface through voice or chat. As an example, a trader could input a free-form query and the EMS would retrieve the data or pull up charts. In the future, we plan to have a large language model (LLM)-based user interface assist with all queries on the dashboard.
The other kind of AI, known as “data AI,” involves building a data-oriented model capable of observing the user’s trading activity and liquidity in the market. This type of AI would act as a trading assistant, make suggestions, and shadow the buy-side trader’s workflow. It would produce dealer recommendations, limit prices, or suggest liquid or tradable look-alike bonds. There are a number of established and emerging vendor products in this space, but we look forward to building some of that in the EMS, given the easy access to aggregated and client differentiated liquidity and protocols.
©Markets Media Europe 2023
