Interest in machine learning is surging in financial services, and the capabilities seem endless. But companies have been slow to figure out how best to implement it in their business practices.
I recently joined a panel discussion for financial executives at Bloomberg’s BuySide Week, where a snap poll of the audience and viewers on Bloomberg terminals found that 32 percent have not thought about how to use machine learning to help direct their investment process, and only 12 percent said they are currently deploying machine learning. Twenty-four percent said they have not yet done so, but would like to at a later time.
During my panel, I noted that NextAngles views this process from a few different perspectives: regulatory pressures, board liability, and how to effectively automate. “What becomes a major issue in financial service is that a lot of data is digitized and in some cases it’s not,” I pointed out.
Those “hurdles and stumbling blocks,” as I called them, mean that a strong focus needs to be placed on how to digitize data from internal and external sources, and then evaluate what it all means. In terms of identifying risk, we are then able to spot patterns a traditional compliance workforce might miss. NextAngles represents a timely solution. Here’s how I put it to the audience:
“What we are hearing from our customers is that they’re looking for exactly what we are offering: A platform solution. What we’re able to do is take a holistic, configurable platform that’s really focused on surveillance. What it essentially does is help them identify risk and manage their objectives. Taking a look at how we can automate some of the data collection and collation of the data. From a risk perspective, from surveys that we have taken, 37 percent of are still using spreadsheets to manage risk.”
I was also asked by the moderator to reflect on the days when companies were first struggling to assess how they could implement machine learning and AI. I recall that during my tenure as CIO of a major global bank, executives were just embarking on a “journey” to see how the data they collected could benefit the business. There were hundreds of applications supporting the product processes for the retail and commercial banks. We evaluated our Internet applications and our financial center application platform to determine how to add business and help generate revenue. We essentially created predictive models for targeting specific products.
Today, reputation risk is top of mind in financial services, because boards are held liable for compliance failures throughout the organization. Take the example of a bank focused on trade finance: keeping track of import and export documentation is a time-intensive, highly manual, and often complicated process where important information can slip through the cracks.
NextAngles’ solution supports the encoding of rules to examine numerous documents to find discrepancies, which are then investigated by an employee.
Another panelist, Mac Steele, director of product at Domino Data Lab, noted that financial services firms are “pretty excited by machine learning.” He cited research from Bloomberg showing that in company earnings transcripts going back five years, machine learning was almost never mentioned, while in the most recent four quarters, “it’s gone parabolic.”
However, he cautioned that excitement is slow to translate into adoption: “It’s hard to separate the hype from reality. A lot of people are confusing big data, AI and machine learning. If you peel back the onion, a lot of people are talking but not enough are doing things with it. It’s hard to build a competitive advantage only on data.”
The third panelist, Gary Kazantsev, head of machine learning engineering at Bloomberg LP argued that while harnessing data to your advantage may be challenging, “it’s your most important lasting advantage. Algorithms are a way to optimize what your objective is, but unless you have something to train those algorithms on, no amount of them are ever going to help you.”
I anticipate that in the next few years the deployment of Machine Learning applications will increase significantly from 12%. The firms that are currently researching this domain will have the potential to disrupt the ecosystem and impact the landscape of financial services.