The Financial Industry Regulatory Authority (Finra) is in the process of expanding its use of machine learning for market surveillance as it continues to refine its algorithms to trace manipulation.

Steve Randich, chief information officer at Finra, tells WatersTechnology that the regulator plans to increase the use of AI for surveillance to handle easier-to-detect instances of fraud, thus freeing up human surveillance professionals to focus on more complex instances.

“We have used machine learning to make sure that the handling and disposition of the alerts has a higher level of certainty in that judgment. We are training the machine to do what the humans do in terms of their initial judgment and intuition and let the algorithm do that. That’s exactly where we’re at,” he says. “Market manipulators are getting smarter so when they notice that they’re being caught they will change tactics. That’s why we still need humans involved. Our plan this year is to continue doing more on the behaviors most commonly used by fraudsters. The roadmap is to implement machine learning so that the human is doing less of the redundant, low-value work of invalidating false positives.”

The regulator uses machine-learning algorithms to detect spoofing and layering activities. Spoofing is when a trader bids on a stock with the intention to cancel before execution, while layering is when a trader has multiple orders but will not execute to create the illusion of liquidity. Randich says algorithms trained to detect other forms of market manipulations will be rigorously tested before being deployed to full production.

Randich notes that Finra sees around 165 million market events in a single day, so the regulator saw it fit to move some operations to the cloud to be more agile, decrease costs and take advantage of the huge amount of data for analytics that the cloud can store.

Finra, which handles cross-market monitoring, historically relied on human judgment to determine if there was potential market manipulation, but Randich points out fraud patterns are now spread among exchanges and trading venues. Human analysts may have a harder time spotting patterns as fast as machines can. These analysts however still monitor the work of the machines to help in recognizing new patterns though the plan is to train an algorithm enough that it can process more information to figure out new manipulation patterns by itself.

During a presentation at this year’s AWS New York Summit, Randich also pointed out that machine-learning algorithms could be applied to the consolidated audit trail (CAT) to quickly identify instances of spoofing. The CAT—which Finra is now currently developing after last year’s ouster of Thesys CAT—will be the largest database of equities and options transactions in the world. During his presentation, he said that the regulator will be looking at several forms of AI to incorporate into the platform that Finra is building.

“Finra is well-positioned to deliver on this system; we started working on it earlier this year,” he said. “So going forward, we plan to leverage machine learning, artificial intelligence, natural language processing, deep learning neural networks to become a more effective and efficient regulator to find market manipulators, fraudsters, insider traders and just bad brokers.”