A journeyman Wall Street technologist believes he has cracked the code to loosening banks' grip on a hugely profitable corner of financial markets.
Miles Kumaresan left his job as global head of trading and fintech at Nordea Asset Management in October because he thinks he's found a better way to enable buying and selling in the $8 trillion market for corporate bonds. It's one of the last major holdouts of the old-school way of trading, where individuals at big banks including J.P. Morgan and Goldman Sachs use phones or chat programs to find bonds for clients on the so-called buyside, including major asset managers like BlackRock and Pimco.
Kumaresan's London-based start-up Wave Labs aims to cut out middlemen who last year pocketed about $9 billion in fees and allow financial firms to trade directly with one another using software.
"I realized how awkward and backwards bond-trading worked" while at Nordea, Kumaresan said. "You always have to go to a guy who has to find it from someone else. If 99.9 percent of the liquidity is with BlackRock and others on the buyside, well we should be able to trade with ourselves. Why pay the middle guy?"
Success by Wave Labs, which launched last month and has three employees, would be an unprecedented coup. Electronic trading is faster and cheaper than so-called voice trading and has taken over everything from stocks to currencies, but the market for big corporate bonds has remained stubbornly human. Some firms have made inroads, but electronic bond trading is still very small scale. About 80 percent of U.S. corporate bond deals by volume are done by phone or chat, according to Greenwich Associates.
That's because there are many more bonds than stocks (General Electric has one stock but more than 1,000 types of bonds, each with different attributes, for instance) and they tend to trade far less frequently. Given those characteristics, the market has remained an informal network of dealers dominated by the biggest U.S. investment banks.
Wall Street has a strong incentive to keep things that way: Banks generated $9 billion in revenue last year from corporate bond trading, according to research firm Coalition. For firms like J.P. Morgan, Goldman and Morgan Stanley, credit trading is part of a suite of services offered to institutional investors. These banks also help companies issue new bonds, and access to these products is one reason clients remain loyal to brokers.
The result is that, apart from inroads by firms including MarketAxess and Tradeweb for small trades, the dynamic hasn't changed much in 30 years, according to Jim McCormick, retired CEO of Chicago-based trading shop TransMarket Group. McCormick is an investor in Wave Labs.
"It's easy to sell $1 million, but God help you if you have to sell $100 million, or buy $100 million," McCormick said. Large orders like that puts funds at the mercy of broker-dealers, and you "hope they don't take advantage of you too badly."
It's not as though people haven't tried to disrupt this market. Over the past two decades, dozens of electronic trading start-ups with names like BondBook and Bondcube — some backed by the banks themselves — have failed to catch on.
But Kumaresan, 54, says he has a novel approach. Born in Sri Lanka and raised in Norway, he is one of five children of an accountant and cook. Chasing his future wife across the globe, Kumaresan got his first job as an IT specialist at an Australian stock broker. He has degrees in mathematics and artificial intelligence and says he is mildly dyslexic.
After becoming obsessed with markets and financial data, he moved from Sydney to positions in Singapore, New York, London and Copenhagen, writing trading algorithms to boost returns at places including Credit Suisse and Dresdner Kleinwort.
The first part of his solution is a workflow tool that allows portfolio managers — the people who make investing decisions — to dictate a complex set of needs to a trader rather than merely requesting the name of a specific bond. In theory, doing that should broaden the possible matches for suitable assets to be bought.
He says his key innovation is an algorithm that he and his partners have written that helps source potential matches on existing electronic venues. Citing concerns that someone will copy him, he would only say that the algo mimics human traders' psychology.
"The good thing about having an outsider look at this is I needed to try to make sense of it for myself," he said. "I never accepted anything as, 'This is just the way it is.' If something doesn't work, let's change it."
Leaning on his quirky charm and the bravado of a true believer, Kumaresan says he has gotten meetings with some of the world's biggest asset managers. He mentions their names — giants in the industry — and then requests that they stay out of print. As he tells it, the demonstrations of his prototype usually end abruptly as executives gush over its potential.
"I have no competition," he declared. "There's not a single technology that is even remotely doing what I am doing."
Should his platform gain acceptance when it's released next year, banks will eventually be forced to cut trading personnel, Kumaresan said. When a market goes electronic, the cost of trading collapses as venues compete to gain the most scale. The ranks of credit traders and salespeople have already been declining for years amid a post-financial crisis slowdown; they have fallen by a third to 3,300 since 2010, according to Coalition.
He faces skepticism from across the industry. According to a former head of global credit trading at a major Wall Street firm, Kumaresan would probably struggle to dislodge other electronic providers, including Tradeweb, MarketAxess and Bloomberg.
"What this guy will have a hard time getting is desktop space," the trader said. "And if it doesn't have critical mass, it won't matter."
Yet electronic trading is likely to succeed eventually in credit markets, even if it is banks that hold onto the market. Firms including Goldman Sachs and Morgan Stanley have said they are working on it. Marty Chavez, Goldman's co-head of trading, has said that steep gains in computing power will ultimately be able to handle the complexity of the corporate bond market.
When that happens, the users of the instruments — big institutional investors including pensions, asset managers and endowments, and the ordinary investors who are their clients — should benefit from lower transaction costs and greater returns.
"Clients are paying intermediaries huge amounts of money in what is effectively a monopoly," Kumaresan said. "It's ultimately moms and dads and old people who are actually paying those trading costs."
Whether he's a visionary or merely a gifted pitchman — or something in between — it won't take long to find out. By April, Kumaresan will conduct live testing with two or three early clients. He said the service will be live by June, when Wave Labs should have seven employees.
"The moment one or two people are out there showing this works, it will skew the edge in their favor," Kumaresan said. "Others will have no choice but to use this technology. Full stop."
Correction: This story was revised to correct the spelling to Kumaresan.