April’s volatility was a major test for fixed-income markets, driven largely by US policy changes. Dwayne Middleton, global head of fixed income trading at T. Rowe Price, speaks to Trader TV about how his desk navigated the heightened volatility, where his traders are leveraging a diverse selection of protocols and unpacks how much of its fixed income flow is now automated—saying “we’re certainly not trading in an automated or electronic fashion just for the sake of it.”
Middleton also shares his views on the buzz around agentic artificial intelligence (Agentic AI), where it’s currently being applied to the trading desk, and how these applications could evolve in the future.
Interview
Jo Gallagher April was a real stress test for fixed income markets and we saw record volumes year to date. How did your trading desk have to adapt its execution strategy?
Dwayne Middleton April’s volatility was really driven by policy action, and it was most impactful across rates and credit. There were pockets of inconsistent liquidity, but what really helped us was the maturation of the ETF ecosystem, which allowed for a lot of price discovery, price visibility. You never really saw panic. Market activity is pretty orderly, and I think the reasoning for that is the diversity of trading protocols across the marketplace really allowed people to be flexible in how they wanted to source risk. Trading volume activity was probably averaging anywhere from 35 to 45 billion depending on the day over the course of the month. And then once you had kind of the tariff reversal, you quickly had to adjust from being defensive to say more offensive really over the course of a couple of weeks. What really helped us across portfolios was embracing macro products, which we’ve been doing for quite a long time. And what I mean by macro products are using portfolio trading, using CDS indices, TRS and if you’re in the rates space, doing more interest rate swaps, for example. That allowed us to tactically capture opportunity or large market swings without kind of changing the underlying portfolio, either on the downside or back on the upside.
Jo Gallagher I’ve heard anecdotally from firms that there was an uptick in the usage of algos during this volatility. Is that what you saw and what is your process for selecting algos?
Dwayne Middleton I wouldn’t say there was an uptick. I think those protocols were already in place. I think the nuance is that if you go back to early COVID, a lot of the dealers and a lot of the algo-driven type strategies were kind of shut down. Just given the pace of widening, the pace of price drops. The algos really couldn’t keep up with the volatility in the marketplace. That really wasn’t the case here. I think, the algo certainly have matured over the last five years. We’ve incorporated automation, electronic trading across rates, global rates, you know, derivative products, CDS indices, for example, certainly large segments of investment grade, even parts of high yield. The number of tickets we’re probably putting anywhere from 75% to 80% of our ticket volume through an automated or electronic trading protocol. In terms of volume, you know, it depends on the asset class. So IG rates are notional volume with those two sectors–it’s probably a combined average of 45% to 50%. I would say large levered cap structures when high yield. Certainly large segments of EM, securitized products are still largely voice-driven. And that’s really just the function of you don’t have the transparency and pricing and relationships still matter in terms of moving large block size in those particular asset classes. We’re certainly not trading in an automated or electronic fashion just for the sake of it. There’s real value in the price capture and price discovery through those trading protocols. And we’re using that information to feedback into our investment process to help us get smarter on some of the larger higher touch trades that we’re doing.
Jo Gallagher Agentic AI is a hot topic right now. What are your thoughts on the technology and where do you see the greatest applications for its use across fixed income?
Dwayne Middleton We’re not using AI-enabled tools to make kind of risk-based decisions. It’s more of a part of the decision-making process to help traders be more educated in the risk transfer process. So using it to help us assess trace pricing anomalies, for example, or running across our holdings and matching that up with dealer axes and dealer inventory so we can do more targeted trade ideas across the investment platform. But also I envision we can start using it you know, identify pockets of liquidity that might not be, you know necessarily visible just from an inventory and axe run, and then kind of incorporating that with our research team so, you know maybe they could be a little bit more proactive on their research ideas. If you’d asked me where the future direction is going, I would say traders being partnered with AI agents to help them make more data-driven decisions. And you know that could be, you knows, news alerts, you know looking at nonlinear risk factors, for example. The recent tariff volatility is an example of that where a lot of investors had never been through a vol period like that that was solely driven by policy action. I think an AI agent that could source different perspectives almost in a real-time fashion while you’re engaged in the market will help you make better data-driven decisions.
Jo Gallagher Thank you Dwayne for your insight and of course you for watching, this has been Trader TV This Week.
To watch other Trader TV This Week shows go to – Video
Or make sure to follow us on LinkedIn – Trader TV’s LinkedIn
Available as a podcast – link
