Compliance Taps AI
At first blush they seem strange bedfellows — compliance, perhaps the dullest function in financial services, and artificial intelligence, the sexy, cutting-edge fabric of the future.
But in fact, as AI gains traction on Wall Street, the compliance department is an early foothold.
“One of the major areas in which AI can play a role is compliance,” said Jos Stoop, director at Sapient Global Markets. “This is because a lot of it revolves around behavior monitoring and trying to understand the ‘why?’.”
Stoop explained there are three key aspects to compliance: opportunity, motivation, and pressure. “Opportunity means determining if something could have happened. Motivation looks at why somebody would do this and whether there any indications as to the reasons. For example, is there a typical pattern that people have followed in the past that we can use to predict if they’ll do it again?,” Stoop said. “Pressure is far harder to determine because the data isn’t often available, but we can for example attempt to recognize somebody’s moods from the type of language they use.”
“AI can help out when you have certain trigger indications that seem out of order,” Stoop added. “When you have determined the ‘what’ of a trigger event through correlation engines, pattern recognition, or history, you can use AI to study why the trigger could be significant, and what non-compliance events could be related.”
Compliance itself has been elevated in its importance in recent years, amid sweeping rulesets such as Dodd-Frank in the U.S. and Emir and MiFID in Europe, targeted initiatives including Know Your Customer and Fatca, as well as provisions against money laundering, insider trading, and collusion.
On the surface, Wall Street’s biggest response has been to add bodies. J.P. Morgan hired 8,000 people in compliance and control functions since the global financial crisis of 2008-2009, HSBC 5,250; last year, Deutsche Bank said it would add 500 compliance staffers, even as the firm cut thousands of jobs firm-wide. ‘Buzz’ at industry conferences over the past couple years has pegged compliance as the only growth area within financial services.
More hands can do the work, and incremental improvements in workflow technology can move the needle on optimizing accuracy and efficiency. But short of financial regulators declaring all rules null and void, only artificial intelligence has game-changing potential in terms of enabling financial companies to check the box on compliance.
By its often monotonous and repetitive nature, and the massive volumes of data and information that fall under its remit, compliance lends itself to machine-based solutions. “There are things that algorithms and programs are really good at doing, much better than humans,” said Eric Crittenden, portfolio manager at Phoenix-based Longboard Asset Management, which deploys machine learning in managing the data of its managed-futures investment strategy.
Crittenden cited long calculations and databasing as examples of “where fatigue and the failings of the human brain become impediments to success. You want to outsource those types of duties to properly constructed forms of synthetic intelligence.”
“Research efforts in compliance are usually behind the curve because compliance is not a profit center,” said Crittenden. “But there is a lot value that can be added by mining the data and setting up intelligent queries and algorithms to go out and look for red flags. There will be some unintended consequences from that, some iterations and some frustrations, but over time you’ll see compliance become more effective and more efficient from the use of synthetic intelligence.”
For providers of AI compliance tools, the prevailing winds are decidedly favorable, as financial firms urgently seek ways to clear a rising regulatory bar while containing costs.
“There’s a lot more desire to use AI these days, and a lot is being invested in AI to build out additional applications and discover more use cases,” Stoop said. “We’re looking at a number of those use cases right now, starting with internally available data for HR, emails, text messages, etc. Massive scale data, such as trade data or market data hold a lot of potential, but hasn’t usually been included because we’re just not there yet. We need to start at a manageable scale and build from there.”
As an indication of the potential of artificial intelligence in compliance, Stoop noted that Facebook is testing how AI can recognize photos, which could lead to newsfeeds more closely tailored to users’ interests.
“It’s not a far stretch to look at that from a compliance perspective,” Stoop said. “For example, you could examine browser histories to see if what they are browsing for points to a sudden life-event change that would make them susceptible to external pressures to attempt something illegally profitable. You could incorporate changes in their LinkedIn profiles to try and discover if things like the connections they’ve made could relate to what they’re doing inside the company.”
Featured image by C
AiPEX with Watson simulates a team of analysts and traders to identify potential investments.
Machine learning models systematically scan newly arriving, anonymized data to identify anomalies.
The Cobalt programme launched in 2018 to help fintechs collaborate with the asset manager.
Users with different skill levels will be able to undertake machine learning and advanced analytics.
Fintech CEO discusses how AI applications have fared in highly volatile fixed income markets.