Bespoke Algorithms Entail Risk
Sell side banks and market makers are urging their customers to exercise care when customizing trading algorithm, and some, like Barclays, have taken a very hands-on approach to algo customization.
“We’re actively engaged in customization. One challenge is to deliver customization in a cost effective and risk managed way,” said Merrell Hora, head of equities algorithmic trading product at Barclays. “It’s common for clients to think they want customization, but when they commit resources to build it they realize they don’t need it. It’s important for us to invest the analyst time up front to work with the client and figure out what they’re trying to accomplish.”
In 2013, Barclays performed 12 large-scale customization products for clients in the U.S. and Europe. “We believe we’re one of only a few banks doing this type of customization, where we go to the client, analyze their order flow, and come up with a unique solution,” said Hora, who is responsible for building the equities agency trading algorithms, from quantitative research to implementation and customization on the client side. “You would think that clients would require similar solutions, but it turns out that the differences in solutions are striking.”
Wolverine Execution Services (WEX), a subsidiary of Chicago-based Wolverine Trading, offers clients multiple venues for execution including a proprietary electronic trading platform, direct floor access, and an agency brokerage trading desk. While it has seen demands for customization rise among some customers, it sounds a cautionary note.
“There will always be clients that want to customize, however, we are seeing a rise in the number of clients that do not want the headaches or overhead that comes with creating their own algos or using algos that require them to set a number of different parameters and conditions,” said Joe Sachetti, partner at WEX. “The marketplace is very complex and requires extensive knowledge, analysis, and testing to create or maintain the performance of an algo.”
WEX was created eight years ago as an independent agency brokerage, utilizing the trading platform that had been built in-house for Wolverine Trading. WEX has its roots in derivatives trading, and it remains WEX’s specialty.
“Our proprietary equity, options, and futures algos do not require client customization and are monitored and optimized real-time by an expert team so that our customers don’t have to bear that burden,” Sachetti said. “They are leveraging our technology, connectivity, research, and the huge investment that we have already made in creating proven algorithms.”
The original attempts at algo customization go back at least to 2007, although then it was done more in a building block manner. In today’s environment, however, banks are under tremendous scrutiny, internally and externally, to make sure that everything they do has been fully vetted, stress tested, documented, and is properly controlled. With the building block approach to customization, that would be prohibitively expensive.
“The building block approach is to have a bunch of tools that you could just slap together like Legos,” Hora said. “We’ve taken a different route. We have the ability to parameterize our algos differently based on what we understand of the customer’s side of the workflow. We can use the same code and just feed it different tuning parameters.”
Hora uses the analogy of a high end sports car. “You don’t typically just buy the car; you have to pick packages, and within each package you have various options,” he said. “Assuming the customer is experienced enough to drive that high end sports car, we are still responsible for the car. Lately, the requests we have been getting from customers is that they don’t want to be bothered with the details. They are asking us to customize it for them. We’re doing intense customizations only for strategic customers; it’s something of a premium service.”
Image via iStock
Algorithms have become more prevalent in the spot FX market.
QB’s Algo Suite for futures market trade execution is also being co-located to HKEX.
Breaking data silos is key to deploying automation beyond 'nuisance' orders.
They can be used on quantum hardware expected to be available in 5 to 10 years.
Streaming blocks change the basis of matching and price discovery so institutions can find new liquidity.