(Bloomberg) -- The Dutch Authority for the Financial Markets is working with the European Securities and Markets Authority and other supervisors to assess how trading firms can improve algorithm pre-trade controls to ensure that erroneous orders do not cause disorderly market conditions.
The controls, brought in under MiFID II, require algo traders to check factors including the price and volume of every order before sending to the market, to prevent inaccuracies creating dramatic swings. Supervisors are collectively working on strengthening the rules, with a report likely to be published at the beginning of next year.
The Dutch regulator, or AFM, has also linked up with the University of Oxford to research computer-driven trading models and assess market manipulation risks, Robert Graumans, a data scientist in the AFM’s multi-asset surveillance team, said in an interview. Initial work that draws on data from Euronext Amsterdam has looked at how different algorithms respond to various triggers, to help supervisors better identify models that might be harmful to market functioning. The findings were published in the Journal of Financial Econometrics this month.
As artificial intelligence advances, algorithmic trading is accelerating across asset classes and regulators are racing to understand how they can best monitor the technology. Academics at the Oxford-Man Institute of Quantitative Finance have found self-learning algorithms to have many unintended consequences, including the ability to collude to maximize profits, even if not built to do so.
Graumans says regulators still have a long way to go when it comes to understanding algorithmic trading and its risks.
“We all have some kind of idea what triggers the trading behavior of humans but we don’t necessarily have a similarly good idea about what triggers the behavior of algorithmic trading firms,” he said. “It’s good for regulators to admit that we still have a lot to learn.”
The AFM’s research partnership with Oxford started in 2021 and the working relationship is continuing to investigate computer-driven trading models and help detect market manipulation.
“We think it’s important to be informed by academia,” he added. “They work on cutting-edge ideas, which we might not even be aware of. They have the time and resources to explore new knowledge, while we might be more hands-on with practical things like policy.”
Graumans joined the Dutch regulator in 2017, after a short stint at the Financial Industry Regulatory Authority, or FINRA, in New York. He started his career as an equity day trader for a small proprietary trading firm at the Amsterdam Stock Exchange.
(Updates to add context on pre-trade controls. An previous version of this story was corrected to clarify the AFM is working with ESMA on pre-trade algorithm controls, not testing.)
©2024 Bloomberg L.P.
