The financial services industry has often been an early adopter of cutting-edge technology, and today there is almost no part of the trading landscape that remains untouched by Artificial Intelligence (AI).

After years of hype, machine learning solutions are finally helping brokers to reimagine the way they interact with their customers and markets. Recent developments in advanced data analytics are particularly profound, having rewritten the way that firms build models, execute trades, and analyze trends to generate investment ideas and build portfolios.

Watching all of this play out with particular interest are the regulators, who are crunching unprecedented amounts of data in their own efforts to police those they supervise, while also helping to shape the boundaries of this exciting new era.

At the US Financial Industry Regulatory Authority (FINRA), the Office of Financial Innovation (OFI) is a unique unit dedicated to understanding and addressing any issues related to new technologies entering the market.

Changing business models

Haimera Workie, head of OFI inside the self-regulatory agency, says there has never been a more thrilling time to be engaged with the visionaries who are driving tomorrow’s financial markets.

“Something really exciting to me about what we do is the ability to engage with firms on how their business models are changing as a result of technology and innovation,” Workie says. “Moreover, the need to have a dialogue regarding the impact of financial innovation is not a one-time static thing and instead is evolving over time. Our Office of Financial Innovation was designed as a recognition of that continual process.”

Last year, conscious of the explosion in the popularity of data analytics and AI across the sector, FINRA went to the industry to ask how it wanted firms to deal with the challenge of regulating this type of innovation. It received a deluge of responses. A detailed and thought-provoking paper was released collating these suggestions, which Workie said would guide FINRA’s future approach.

“Firms are pleased that we are engaging with the industry on AI, but think specific regulation regarding a specific type of technology is premature, especially as it is developing.”

Haimera Workie, head of OFI

“Firms are pleased that we are engaging with the industry on AI, but think specific regulation regarding a specific type of technology is premature, especially as it is developing,” he says.

Broker-dealers also indicated that they valued the time spent with their regulator in addressing these developments and appreciate that regulators are willing to listen.

“There were discussions about difficult areas, such as AI explainability, model management capabilities, ethical AI, and record keeping. Firms are working through, or in some cases struggling with, some of these issues,” Workie says.

There are three areas where AI has been utilized that are discussed in the white paper:

• communication with customers;
• investment process;
• operational functions.

“Firms are increasingly looking to use AI to enhance their customer’s experience, with tools such as virtual assistants and algorithms that target outreach to existing or new customers,” Workie says.

“For example, firms may facilitate the ability to find out account balances by speaking with an in-home virtual assistant device, or may use website view data related to a client to help target pertinent information to the client.”

“We’re seeing AI in the pre-trade process where you have social media sentiment data that is gathered, and fed into an AI-based algorithm to find out how social sentiment was potentially impacting various stocks, which is then used by market participants to help make investment decisions,” Workie says.

“On the trading side, we see a number of firms have or are exploring the development of AI processes to facilitate trading activity; these systems typically utilize machine learning designed to optimize trading activity.”

In Compliance, Workie says that there are a huge number of useful AI applications appearing, with the largest number focused on anti-money-laundering/know-your-customer applications.

“It needs to be understood there are a lot of potential benefits presented by AI and other forms of innovation; innovation can open so many opportunities for investors and for firms operating in the securities industry,” he says. “At the same time, both industry and regulators have to be cognizant to think through whether any new risks are being introduced, to make sure there aren’t any unintended consequences.”

Global guidelines

To achieve this, FINRA works closely with government partners in the US, such as the Securities and Exchange Commission, but also across the globe, collaborating with the UK Financial Conduct Authority, and the International Organization of Securities Commissions (IOSCO), which sets global guidelines.

“There is a real desire to have cooperation and collaboration in AI, both among regulators, and between industry and regulators,” he says. “A lot of large firms that work across borders have very similar systems – so they want to limit the potential for any regulatory requirements in some jurisdictions that don’t square with the rules in others. Collaboration among regulators should hopefully make it simpler for these firms to develop a systematic AI approach.”

One of the hot topics currently being picked over by FINRA’s OFI is the issue of data bias that may exist within the use of AI directly, and specifically identifying the potential for this to affect the results achieved.

“Any systems that may have biases reflected in the data is something firms should be thinking about when using or training AI applications,” Workie says. “New technologies don’t make the older requirements that exist go away.”

A survey of financial institutions conducted by the Bank of England among a third of banks in the UK concluded that the coronavirus pandemic has impacted the performance of machine learning models negatively.

Systemic shocks

The central bank has previously found that existing risks may be amplified, or indeed new risks may have emerged, from the use of machine learning and data science in financial services. The survey found that some models performed poorly when applied to an anomalous situation that they have not encountered before in their training data.

These findings were shared with regulators around the world, and the issues raised go to the heart of the OFI’s mission. For regulators, the trick is finding the right balance between encouraging firms to innovate freely, and applying regulatory supervision that might stymie development.

Insufficient AI oversight may increase the risk of systemic shocks, while overzealous rulemaking could have the adverse impact of making it even harder for smaller firms to break into the market. Both concerns are carefully considered by Workie’s team as they go about their task.

“Since the formation of FINRA’s Office of Financial Innovation, I have tried to foster an openness to innovation and the opportunities innovation present, while also encouraging everyone to think through the implications of any innovation for investor protection and market integrity before you jump in,” he says.

“There is a danger in not being open to innovation and not being open to what the possibilities are, but not looking at the risk or implications carries a danger too.”