Bond Traders Need to Up Their Game as AI Systems Get Smarter

(Bloomberg) -- Money is pouring into artificial intelligence in bond markets, challenging bankers and investors to adapt their skills in everything from issuing to trading securities.

Fintech startup Nivaura is investing in technology to automate debt sales. Dutch bank ING Groep NV is improving systems to help traders buy and sell bonds, while AllianceBernstein Holding LP advanced its virtual assistant to identify notes that people miss.

After taking over stocks, computers are slowly overcoming resistance in one of the most technology-averse corners of financial markets. Bond traders are wary of a one-size-fits-all approach coming from equity markets, which are now largely automated. They say that human relationships are at the center of the market and clients want to talk through complex transactions.

“More and more aspects of the job and of the investment process will be automated, but it’s not a big bang,” said Santiago Braje, global head of credit trading at ING, who will discuss AI at the International Capital Market Association’s conference in Stockholm this week. “It’s going to be a gradual transition.”

High Touch

AI has proved particularly useful in replacing manual tasks such as inputting data and executing small, liquid trades in markets such as currencies. It’s only just beginning in areas like corporate bonds, that traders call “high touch” for the traditional level of human involvement.

Still, proponents say tech is being used as a tool by people rather than a replacement for them and that it helps firms use human resources more efficiently.

“If you close your mind to it and assume they’ll take your job, you’re missing the point,” said Chris Perryman, senior vice president of fixed-income trading at PineBridge Investments which oversees about $93 billion and uses algorithms to trade liquid securities. “The tasks that humans perform are changing and you need to have human resources allocated to tasks that add value.”

Human Skill

As asset managers increasingly seek an edge through AI and data, demand for human skills is shifting. In some cases, it’s increasing rather than reducing jobs -- as seen earlier this month when Janus Henderson Investors said it would replace two fund managers with a quantitative team of four.

In primary markets, London-based Nivaura expects to hire more engineers and data scientists to automate processes like producing detailed prospectuses for debt issuance, according to Chief Executive Officer Avtar Sehra. The company closed $20 million in seed funding earlier this year and created the first regulated cryptocurrency bond in 2017.

“Complete automation is where we’re trying to get to but we’re far from that right now,” said Sehra, who will also speak at ICMA’s conference in Stockholm on Friday. “Flying a plane on autopilot is actually a lot more of a constrained problem than trying to identify patterns in the market for pricing deals.”

AllianceBernstein, which manages $555 billion of assets, has combined different electronic systems to create a virtual assistant, Abbie, that tells its investment managers and traders which bonds to buy and sell based on pricing, ease-of-trading and risk. The firm says that Abbie shows how humans and machines can work together.

Banks Adapt

Banks are also looking to adapt. Citigroup Inc. created a fintech division in September to invest in technologies related to trading infrastructure, data analysis, artificial intelligence and machine learning in debt markets.

Spanish lender Banco Bilbao Vizcaya Argentaria SA has created a unit to automate trade processes and generate intelligence from data, reducing differences between markets such as fixed income, currencies and equities.

For ING, AI has helped improve bond-trading performance. The Dutch bank said it won 20% more market-making transactions and reduced costs in its first year of using software that suggests easy-to-trade bonds.

It has since expanded the platform, named Katana after a samurai sword, into a tool for asset managers, and it’s finessing algorithms and adding trade data to increase speed and focus. Such systems help encourage electronic trading of corporate bonds in larger sizes -- the type of trades that are still usually done by phone and instant message, Braje said.

Bloomberg LP, the parent of Bloomberg News, offers bond-trading services to banks and fund managers and provides services that facilitate ordering and distributes information on new debt offerings.

“New developments are focused on analytics rather than data collection,” Braje said. “We’ve seen the first wave and we’re now seeing more and more intelligence being added.”

To contact the reporter on this story: Katie Linsell in London at klinsell@bloomberg.net

To contact the editors responsible for this story: Vivianne Rodrigues at vrodrigues3@bloomberg.net, Abigail Moses, Chris Vellacott

For more articles like this, please visit us at bloomberg.com

©2019 Bloomberg L.P.