Machines oust humans in the market for ‘odd lots’ of corporate debt
Source: Financial Times
A quiet but important evolution is unfolding in a less glamorous corner of the US corporate bond market, as machines oust humans as traders of smaller “odd lots”of debt bought by both institutional and retail investors.
Banks, asset managers, exchange groups and financial technology companies are seeking to replicate in the fixed-income universe the advance of electronic trading that has transformed the equity market in recent decades.
“It’s a massive analogue-to-digital conversion that’s in its really early days,” Jeffrey Sprecher, the chief executive of Intercontinental Exchange, said at a conference this month.
Historically odd lots, which are less than $1m in size and quoted at a discount to larger, more marketable amounts of bonds, were treated by banks in the same way as bigger and more lucrative trades. Salespeople worked with a trading desk to generate a price for clients. But as banks prioritised bigger trades, investors complained that requests for prices in smaller transactions increasingly went unanswered.
Now, oddlots are proving a particularly fertile ground for electronic trading as banks have developed their own algorithms to price them. In the past year, Market Axess — by far the largest electronic bond trading platform with 86 per cent market share by dollar volume, according to Greenwich Associates — has seen the number of banks using algos to price odd lots double from four to eight. Bank responses to investor requests for price quotes via an algo have surged from 287,000 in the first quarter of 2017 to over 653,000 for the same period this year.
“It’s better for dealers and investors,” said Rick McVey, chief executive of Market Axess. “Investors are getting a better price experience for small trades than ever before and dealers are getting in the middle of more trades at lower cost.”
The market consists of small retail clients trading through financial advisers alongside big asset managers and hedge funds dealing directly with banks. MarketAxess, which only deals with institutional clients, accounts for roughly 35 per cent of the dollar volume of all US corporate bond oddlots.
Oddlots accumulate in the market as fixed-income asset managers rebalance their portfolios against a benchmark every month. They are also created when asset managers buy a large amount of a company bond and then distribute it across their individual client accounts.
Credit Suisse, Morgan Stanley or Goldman Sachs are among the banks that have developed algorithms for the market.
Goldman Sachs quietly rolled out its own electronic market-making programme for US corporate debt in 2015, initially only quoting a few thousand bonds and executing trades of $500,000 or less. Today, it makes markets in 10,000 US corporate bonds and handles trades of up to $2m.
Julian Pomfret-Pudelsky, a fixed income algorithmic trader at Credit Suisse said: “For a single trader it takes as long to price a $10,000 trade as it does a $10m trade. It’s very cumbersome.”
By responding to investors that want to trade bonds using an algo, banks can cut costs, while investors say buying and selling debt has become easier, with better prices offered by more lenders, and a narrowing of the traditional discount in the oddlot market of just one basis point compared to larger trades.
The improvement in the ease of buying and selling odd lots is also illustrated by Greenwich Associates’ survey of asset managers on corporate bond trading.
In 2016 only 29 per cent said executing trades below $1m were “extremely easy” and 6 cent gave the top grade for trades of $1m-$5m. Last year that jumped to 51 per cent and 18 per cent respectively. Moreover, the percentage of respondents that said mid-sized trades of $1m to $5m were relatively easy more than doubled to 58 per cent.
“Liquidity has improved,” said Kevin Giddis, head of fixed income at Raymond James. “We do a lot more electronically now. Trading floors aren’t as loud. People are mostly just manning a mouse, not a phone, looking at algorithms scraping liquidity, just like equities.”
There are still obstacles to a complete overhaul of the trading architecture of the fixed-income market. Unlike the highly electronically traded equity market where there is typically just one equity security for a company, a larger company has many bonds with different maturities and yields.
Still, the trail being blazed by algos transacting odd lots suggests that the bond market is set to follow the advance of electronic trading in equity trading. Bond traders see scope for slicing big corporate bond trades into smaller chunks, creating the opportunity to sell them algorithmically in bursts, rather than calling up or messaging big banks for quotes — something that already happens in the equity and government bond market. That would, in turn, accelerate the corporate bond market’s adoption of electronic trading.
“In a world where people are splitting up trades, if you’re not equipped with a full-fledged algorithmic platform you cannot survive,” predicts the global head of investment banking at one bank.
