Algorithms Moving into the Bond Markets

Algorithms Moving into the Bond Markets

Algorithmic trading may fairly be said to have conquered the public equities world, although there are still pockets of resistance and related controversies. The robots have now turned their attention to the bond markets.

Bond markets are different from stock markets in a lot of ways, and many of these ways have converged to slow the advent of the algos in the former domain. There is, for example, the simple fact that a corporation may have only one listed stock, but dozens of unique bonds. There are fewer than 50,000 stocks in the world, in contrast to millions of bonds, and as a writer at the Financial Times put it ten months ago, each of these millions has its own “legal and financial idiosyncrasies.”

Liquidity and ETFs

Still, there has been some movement, and bond algorithms were a subject of discussion at the TABB Group’s 2019 Fixed Income Summit in February. Larry Tabb moderated a panel on bond algos. As Tabb put it, the key issue is whether there is enough liquidity on the credit side for the algos to function.

There is a close tie between this issue of liquidity and the availability of exchange-traded funds. Bonds that are the components of ETFs are much more liquid than those that are not, due to the standardization involved as well as the create/redeem process, and the bid-offer spread is substantially compressed. HYG (a high-yield bond ETF issued by BlackRock) and LQD (its investment-grade companion) were mentioned in particular as contributing to this compression.

“In that sense, ETFs are creating a more efficient [bond] market,” and one more friendly for algos, said Tabb.

A buy-side executive on that panel said that the market is in the early stages of bond algo trading, but this executive’s firms has set up an algo that automatically responds to RFQs, “with certain parameters, such as which part of the curve do you like, which bonds do you prefer, which bonds are attractive, which names are very active by our investment management teams.”

Banks and Non-Banks

The volume of bank responses to investor RFQs by way of an algorithm has been surging. In the first quarter of 2017, that volume was 287,000. By the first quarter of last year, 2018, it was 653,000.

A sell-side expert on the panel said that his bank’s corporate bond algo now prices 10,000 CUSIPs, some investment grade and some high yield. It processes inputs, develops a price, and does some asset management, to “take the burden away from round-lot voice traders” who would otherwise have to respond to all the RFQs coming through electronically.

There are also non-bank liquidity providers in the bond market that are looking for ways to incorporate low-touch, automated investment-grade credit-trading. The TABB panel included the CEO of a proprietary trading firm who said the firm has reworked the architecture of trading workflows, making them both more systematic and more quantitative. His firm, which launched just two years ago, uses a variety of protocols and venues and attracts clients who are delighted by the immediacy of click-to-trade.

This CEO said that 80% of the firm’s workflow is now automated and the goal is to go further and automate “the whole lifecycle of data modelling, processing, data acquisition, portfolio allocation, and risk management.”

Algorithms are also valuable as a pricing tool in portfolio trading. Over time, institutions are bringing bond algo trading within their comfort zone.

The Intercontinental Exchange

In a related development, the Intercontinental Exchange last month created a new fixed-income trading platform, ICE Bonds, that combines three older entities: ICE BondPoint; TMC Bonds; and ICE Credit Trade. The platform will include continuous and end-of-day pricing and analytics by ICE Data Services.

Marshall Nicholson (formerly the managing director of BondPoint) will preside over ICE Bonds, and Tom Vales, who had been running TMC Bonds, will be COO, reporting to Nicholson.

At a conference in June the chief executive of ICE, Jeffrey Sprecher, said that as computerized bond trading moves into the odd lot space, he is seeing a “massive analog-to-digital conversion that’s in its really early days.”

Be Sociable, Share!

Leave A Reply

← Bayesian Probability Theory and a Hierarchical Learning Portfolio Hedge Funds: Observing the Unobserved Performance →