Few phrases have the buzzword cachet of artificial intelligence (AI) at present. It is, literally, the talk of the town at any conference. But while interest is high, few areas of trading have made as much use of it as market surveillance. And out of those, few companies are as advanced in their deployment of AI and its subsets as Nasdaq.
In its sell-side endeavors, the exchange operator has put machine learning technology into production on its Nordic exchanges and sold it to a number of other bourses. For the buy side, it has the Nasdaq Buy-side Compliance (NBSC) platform. Detailed fully in the write-up for its market surveillance platform category win for the same technology (see page 48), NBSC goes beyond most market-oversight platforms by allowing users to assess the behavioral aspects of trading by portfolio managers, to gain a truer picture of what is genuine activity, such as hedging, and what might be market abuse.
Much of this stems from Nasdaq’s 2017 acquisition of Sybenetix, and the incorporation of that company’s Compass platform into NBSC—a stated ambition for the acquisition was for Nasdaq to strengthen its push onto buy-side desks. But the beauty of the platform is in how it meshes traditional high-technology tasks, such as analyzing data, with the fluid and previously unmanageable activity of humans.
Using Function Resonance Analysis Methods, the system is able to determine cognitive causality for decision-making by fund managers, which is expressed through heatscoring to identify behaviors that are outside of the norm. For instance, if a particular portfolio manager is consistently beating the market, yet his peers using similar strategies and trading similar instruments are not, it can compare behavioral patterns in the past to determine if potential abuse is taking place. Similar logic can be applied to organizations and accounts to analyze trading behaviors and determine whether a case warrants further investigation.
“We were able to find behaviors that I would call in the gray zone, which, depending on how you set your alerts, you may or may not detect. Now with machine intelligence, that gives you much better coverage,” says Valerie Bannert-Thurner, head of risk and surveillance solutions at Nasdaq.
Outside of pure surveillance, Nasdaq’s Fair Investor Allocation Reporting module also allows firms to process transaction information for best-execution auditing, and for proving that assets are being allocated fairly and equally. Using machine learning, this can be done far faster than any human analyst could manage and could potentially alert compliance officers to unfair treatment of client accounts before it becomes a problem.
Bloomberg’s Gerard Francis looks at the challenges that capital markets firms face when trying to incorporate alternative datasets.Subscribe to Weekly Wrap emails