Will Robots Save Us From Robots?

The market, it seems, is ready to fight fire with fire.

As electronic trading continues to get more advanced, so too does the complexity around properly surveilling it.

It takes months—often years—to properly digest market events. Trades are pieced together during exhaustively long processes that not only take up a lot of time, but also are extremely costly.

So it was only a matter of time before the decision was made to start using the technology that created such a complicated place to help understand it.

Turning to the Robots

By using artificial intelligence to monitor their firms' trading, they could potentially free up staff in their compliance department to work on the laundry list of other tasks that need to be addressed.

On Tuesday, a story published by Reuters elaborated on that very theory. Nasdaq, the London Stock Exchange Group (LSEG) and the Financial Industry Regulatory Authority (Finra) are all investigating ways to use artificial intelligence for market surveillance, according to the story. The story stated Finra was looking to test the technology next year while the two exchange operators were hoping to use it by the end of the year.

So how capable are these technologies of sorting through the convoluted markets?

According to the story, an investigation into the manipulation of foreign exchange (FX) rates that took over two years to discover and rule on could have been found in real-time with the software Nasdaq is using from Digital Reasoning, the cognitive-computing vendor Nasdaq announced a collaboration with in February.

For more on artificial intelligence and its use cases in the capital markets, read Anthony Malakian's features in the March issue of Waters.

Tom Gira, executive vice president for market regulation at Finra, was quoted in the story discussing how artificial intelligence could even uncover techniques currently unknown to regulators.

"The biggest concern we have is that there is some manipulative scheme that we are not even aware of. It seems like these tools have the potential to give us a better window into the market for those types of scenarios," he said.

Easing the Process

When the topic of artificial intelligence comes up it's often tied directly to the front office and trading. Although they might not talk about it publicly the way they discuss their blockchain projects, rest assured firms are interested in using artificial intelligence to help their traders make as much money as possible.

That's why I think it's interesting to see firms turning to AI for market surveillance. It's outside some of the traditional use cases many had the technology pegged for initially.

Granted, it's not as if these firms aren't also looking to make money from these endeavors. Nasdaq and LSEG both admitted in the Reuters' story they had plans to sell the solutions to banks and fund managers to help them to monitor their traders.

That could seemingly be a lucrative business, as firms are always keen to find efficient ways to remain compliant. By using artificial intelligence to monitor their firms' trading, they could potentially free up staff in their compliance department to work on the laundry list of other tasks that need to be addressed.

From a regulatory perspective, if market events can be deciphered faster, that would also be a big win for the industry.

WatersTechnology US Editor Anthony Malakian and I discussed Regulation Automated Trading (Reg AT) on last week's Waters Wavelength podcast, which has caused an uproar in the industry due to its request for firms' source code. One of the reasons the Commodity Futures Trading Commission (CFTC) is pushing for a source code repository is because it needs to dissect market events. The use of artificial intelligence could significantly cut down on the time spent digesting those events, lessening the need for the polarizing database.

But, to be clear, pure artificial intelligence isn't the golden ticket. Even Digital Reasoning's president, Rob Metcalf, admitted so while speaking to WatersTechnology for a feature on market surveillance last year.

"There's no substitute for having someone with subject-matter expertise in a particular domain," Metcalf said. "Technology enables you to bring in a subject-matter expert, someone who really understands foreign exchange (FX) language, for example, to help train models that are more effective at deciphering and understanding the communication you want to analyze in a foreign exchange (FX) chat. So the trick is combining cutting-edge technology with human expertise."