“Fixed income is so illiquid. You have to talk to each trader at J.P. Morgan or Merrill Lynch until you have spoken with every investment bank,” said Jeremy Son, a fixed-income trader at AllianceBernstein (AB) in Hong Kong. “I have more than a hundred chatrooms with various dealers and a lot of information gets lost.”
Whereas trading and execution in equities and foreign exchange takes place across digital channels, trading bonds is still done manually over the phone or in instant-messaging conversations. The advent of trading tools using artificial intelligence and machine learning, prevalent in those other markets, is still in its infancy when it comes to bonds.
While the largest, most liquid market such as U.S. Treasuries may have become largely electronic, that is not true for other segments – notably emerging-market credit and local-currency bonds.
Indeed, there are few statistics available in Asia, which remains a backwater when it comes to trade-tech. “No one knows what the split between voice and electronic trading is in Asia,” says Jesper Bruun-Olsen, head of Asia Pacific at Algomi, a software company that helps fixed-income fund managers aggregate data.
Liquidity aggregators
That’s because bonds trade over-the-counter, not on an exchange; in a big national market like the U.S., liquidity is less of a challenge, but corporate and high-yield bonds are still often voice-traded. In Asia, each market is small or information is hard to access, so fragmentation makes it impossible to generate liquidity, which in turn means there’s little data available to feed a trading algorithm.
This is less of a problem for small trades below $1 million: that universe already trades electronically. But posting larger trades, especially anything above $5 million, on an electronic platform is asking for trouble. You’re going to get front-run.
Fixed income is so illiquid
Jememy Son, AllianceBernstein
Things are changing, however, as a result of both the external trading environment as well as technological progress. Both individual buy sides such as AB and vendors marketing platforms, such as Bloomberg, MarketAxess and TradeWeb, are racing to create automated, direct relationships with other market participants. In effect, they are using technology to recreate the pre-GFC broker-dealer market structure.
Sourcing ALFA
In 2015, AB’s engineers developed a product called Automatic Liquidity Filtering Analytics, or ALFA. By drawing upon data from various channels including Bloomberg, Neptune, major dealers, inter-dealer brokers, it helped traders scour the market to see if other players had bids or offers out for bonds that AB’s portfolio managers wanted to buy or sell. AB sold the software to Algomi to commercialize it.
With ALFA gathering data from the market, AB then developed Abbie – a virtual assistant – on top of it to pick the best bonds and build orders. This lets a portfolio manager determine whether a bond is liquid enough to fit her desired portfolio, while saving time for traders trying to fulfill her instructions.
Son says AB is working on two enhancements. First is a momentum indicator that guides a trader’s execution tactics. It is already in use for investment grade and high yield bond in U.S.
Second is an execution-management system to let the firm’s bots transact directly in the market. It will be rolled out within three to six months, “perhaps even sooner”, said Son.
Traders like Son used to check upwards of 20 dealers to get a particular bond, week after week if need be. The more thorough the sweep, the more likely word would leak into the market that he was on the hunt, and prices would move against him.
ALFA saved him a lot of bother. But he still has to go to a vendor platform to actually execute the deal.
AB is now building a model to allow the transaction to be done bilaterally without needing to go through an intermediary platform. It is not the only buy side working on in-house, B2B solutions.
All to All
Centralized platforms such as MarketAxess’s ‘all-to-all’ open trading platform are anonymous, but investors are wary about sending large sized inquiries, because that can potentially impact the market.
On the other side, dealers ( mainly investment banks ) are only willing to share information about their axes with “selected buy-side real-money investors like us,” Son said.
The firm is working with sell-side dealers to integrate directly, so that an indicative offer can be transacted without the price being refreshed in the meantime. This, of course, assumes that banks can offer real-time prices.
Some banks are beginning to do so via Symphony, the capital-markets communication platform. Symphony and Bloomberg are the leading venues for marketplace chatrooms, and AB wants to use Symphony as the conduit to banks able to back up direct transaction.
Platform providers: centralizing liquidity
Platform vendors are also racing to overcome bond-market fragmentation. The more liquidity, the more electronic trading; the more trading, the better to prevent information leakage.
Craig McLeod, London-based head of emerging-market product management at MarketAxess, says liquidity has already improved a lot since the 2008 Global Financial Crisis. Before then, the major global banks owned 80% of secondary bond markets. Outstanding notionals in bonds have mushroomed, in part due to quantitive easing and other balance-sheet expansions, but traditional broker-dealers have pulled back from secondary markets, due to post-GFC rules restricting capital. So all of that expanded balance sheet inventory has moved to major banks from Asia and the Middle East.
MarketAxess has an open trading framework that allows broker-dealers within its walled garden to transact, so a trader at, say, First Abu Dhabi Bank can trade electronically with ICBC.
“Asia is predominantly voice traded, but the rest of the world has woken up to the need to be invested here,” McLeod said. More buy sides are putting bond-trading desks in the region, which have to conform to European regulations around best execution. So they are keen to find electronic solutions, and platform providers like MarketAxess have a readymade venue.
“It’s not going to be a Big Bang like you had in the equities world,” McLeod said, given fixed income’s OTC nature. “Clients aren’t getting rid of voice trading but there’s more discussion around e-trading.”
From data come the tools
With electronic trading, and the data underpinning it, comes new applications for A.I. MarketAxess, for example, is building solutions for private bankers, which would give them the kind of alerts and liquidity analysis that was previously available only to broker-dealers.
“The liquidity pool is becoming one,” he said of the vendor’s all-to-all marketplace. MarketAxess is developing A.I. tools to match risk by valuing bond prices, with information sourced from client blotters. It also serves as the central counterparty between the buyer and seller, settling trades. It’s not quite a clearinghouse (which, in a futures market, would require margining and guarantee trades) but it’s a step towards the kind of centralized marketplace that exists in the equities world.
By aggregating more liquidity, and using A.I. tools, the vendor can better protect buy sides against information leakage. Each client has a profile that is tagged; the system keeps track of who honors bids and offers and their advertized prices. The idea is that users can see which accounts are honest players and which are not.
Solutions such as these are eating into the traditional inter-dealer business. But banks are constrained by regulation, and they’ll access risk via MarketAxess or another vendor if it gets them the liquidity they need, particularly in emerging markets. “We want to keep the flow monster going,” McLeod said.
A.I. in the driver’s seat?
Yet these platforms also have their limits. For fragmented markets, the likelihood of a trader pinging a counter that matches – the right bonds, at the right volume, at the right price – is low. Vendors are using alerts and other tools to help, but the majority of volumes will have to remain voice-traded, says Bruun-Olsen at Algomi.
Before big buy sides can start using A.I. and machine learning for notifications, chatbot alerts and so on, they need the data. It’s big data that creates efficiencies. But that becomes a problem in Asia because of its fragmented markets. Traders like to say that ‘liquidity begets liquidity’, but it also begets the data required to make A.I. useful. By itself, A.I. doesn’t beget anything.
So even as firms like AB develop tools to guide traders’ tactics, source liquidity and measure performance, we are a long way from A.I. taking over fixed-income execution.
