One year after launching a credit algorithmic trading platform for investment grade corporate bonds, Credit Suisse has extended its scope to high yield, bringing some 1,000 dollar-denominated bonds into the orbit of CSLiveEx. Over time it could bring the same technology to the European market.
Algorithmic trading came to the bond markets later than other asset classes, only really developing in the last four years or so. Corporate bonds in particular are an area where voice has stayed dominant, a reflection of their complexity compared to equities or currencies, or even government debt. But this dominance varies hugely according to size. In the investment grade corporate bond space, only about 20% of the market by volume is electronic; by number of tickets, it is more like 98%.
It's no mystery why electronic is the preserve of small trades: voice broking is still the preferred way of shifting a big position without blasting your intentions out to the broader market. But that volume/ticket discrepancy also highlights the huge inefficiency at the smaller end of the market: after all, it takes much the same time and effort for a credit desk to price a $5 million trade as it does to execute a deal one-tenth of that size. Banks would much rather spend their human capital on the kinds of trades that make up the bulk of fees, so in the absence of automated systems, most small orders will tend to get ignored.
It was this kind of thinking that drove Credit Suisse to set up its investment grade credit algorithmic platform in late 2017. Goldman Sachs and Morgan Stanley have rival systems.
In a rare example of buyside and sellside interests being aligned, a drive for greater efficiency at the small end is increasingly being demanded by investors. Credit Suisse gets a daily reminder of this from its private bank, which supplies thousands of credit trading orders in the sub-$1 million area and is constantly under pressure from clients to execute them as efficiently as possible.
"We don't view this as an odd-lot business – we view it as electronic."- Julian Pomfret-Pudelsky
The first step – the launch of CSLiveEx in September 2017 – was to build an algo trading platform to cover some 6,500 dollar-denominated investment grade corporate bonds. The system provides firm and executable pricing to MarketAxess, Tradeweb and Bloomberg. For investment grade bonds there is a cut-off of $1 million notional, falling to $500,000 for high yield. Deals of these sizes – sometimes called "odd-lots" – can be from retail clients but are frequently institutional business as funds rebalance portfolios.
“Historically, there has been a cut-off in investment grade at $1 million for business to be considered round-lot versus odd-lot," says Julian Pomfret-Pudelsky, who has been at Credit Suisse since 2013 but has just been appointed as global head of credit algo trading, alongside the expansion of the platform into high yield. "In any case, we don’t view this as an odd-lot business – we view it as electronic, and those limits will change as more and more clients execute electronically in larger size.”
Moving into high yield is a logical enough expansion for a firm like Credit Suisse, given its leveraged finance pedigree. But that doesn't make it a straightforward development.
"It's a natural step for us, but the secondary market in high yield is less electronic than investment grade – perhaps 10% or 11% of volume," says Pomfret-Pudelsky. "We do see more demand from customers and we do expect that to increase further. However, high yield is a more complex and less liquid product. The value-add of human interaction increases as you move down the credit spectrum."
The challenge, then, is to make sure that the pricing generated by the platform is as accurate as possible even with these obstacles. It's partly why the trade size cut-off is half that of the investment grade platform.
"There is a lot of research that goes into building the CSLiveEx trading strategy – it needs to be able to do what a good trader can do," adds Pomfret-Pudelsky. "But ultimately, there is only so much information that we can process electronically. So the challenge is finding a balance between the mathematics of the strategy and the technicalities of the market."