May 20, 2022
Credit Algos Aren’t Sweating Market Stress Like They Did in 2020
(Bloomberg) -- When the onset of the coronavirus pandemic roiled financial markets just over two years ago, credit trading algorithms were flummoxed. In many cases, dealers turned their algos off and handled corporate bond trading the old fashioned way -- over the phone and through instant messages.
Now, traders say robots are better equipped to price bonds during periods of market stress, in part because of data gathered during the 2020 turmoil. With volatility ramping up quickly as the Federal Reserve tightens financial conditions, new and improved algos are again being put to the test -- and so far they’re passing.
“Algos and automated trading have made a lot of progress over the last few years,” Mike Sobel, chief executive officer of electronic bond trading platform Trumid Financial LLC, said Wednesday during a panel at Bloomberg Intelligence’s Market Structure 3.0 conference in New York. “It’s not just going to go away because vol picked up, which, in previous iterations, we took a big pause and maybe a step backward. I don’t foresee that happening this time.”
The more data an algorithm has to work with, the better it will perform. The market stress of early 2020 -- when Treasury yields swung wildly and credit spreads widened so far that new company bond sales froze entirely -- has now proven to be a valuable input for trading algorithms that previously hadn’t operated in an environment that volatile.
Broker dealers have also had time to tinker with and improve their electronic credit trading technology and strategies.
“If we rewind a few years ago, the majority of algos out there were not quite mature. Now, the algos are a bit more mature and we all have better tooling around how to manage our algos during times of stress,” Justin Lada, global head of electronic credit strategy at Barclays Plc, said in an interview. “I think our ability to react during times of stress and not just turn them off is better now than a few years ago.”
When corporate bond prices become unpredictable and buying from clients could lead to quick losses, dealers tend to pull back from the market. One way to do that is by shutting down automated trading, especially if the robots are spitting out prices that look wrong. As that tech improves and algos can stay running, theoretically liquidity should, too.
After the initial spike in Covid volatility took some algorithms offline, electronic and algorithmic credit trading accelerated as markets calmed -- which became a necessity since investors and traders were working from home. In April, 40% of trading in the US investment-grade market was fully electronic, according to research firm Coalition Greenwich. That’s up from 29% in May 2020.
Sonali Theisen, Bank of America Corp.’s head of fixed-income e-trading and market structure, said during Wednesday’s panel that that may have been a turning point for the market. “That was actually the first time during a rise in volatility that we actually didn’t see the proportion of electronic trading go down,” she said. “It held steady and, in some cases, it actually went up. I think that’s a harbinger of things to come.”
While corporate bond prices have dropped sharply this year, the market has not yet been stressed to the degree it was in 2020. But lessons continue to be learned that should help during the next panic.
“Things will be tested now, and that’s good,” said Trumid’s Sobel.
©2022 Bloomberg L.P.