Technology is radically changing the way we learn. 

There is evidence that the rapid evolution of how humans interact with technology is rewiring our brains. In 2015, Microsoft researchers in Canada surveyed 2,000 participants and studied the brain activity of 112 others using electroencephalograms. “Tech adoption and social media usage are training consumers to become better at processing and encoding information through short bursts of high attention,” the report said, and researchers also found humans are improving at focusing on more things at once, concluding that human attention is becoming more discerning and demanding of quality content to consume. 

Educators are adapting their methodologies to adjust to evolving cognitive behaviors by deploying new technological tools designed to teach, taking advantage of new technology that fosters collaboration, makes geography irrelevant, and offers powerful visual and immersive opportunities to absorb information. 

It follows, then, that the way knowledge is passed on within financial services is also undergoing a transformation. Firms and universities are entering creative, mutually beneficial partnerships that make traditional internships look positively quaint, building programs that embed students into firms to solve tricky problems. 

NEET Idea

The Financial Risk Group (FRG), a risk management consultancy, runs a program that was relatively pioneering when it launched in 2009. 

“The concept isn’t exactly new,” says FRG managing partner John Bell. “Folks have been apprentices for hundreds of years, but in the tech space, it seemed like a fairly novel concept for some reason. We didn’t have a lot to draw on back then.”

New Employee Excellence Training (NEET) was created because FRG “had a heck of a time finding anybody to hire,” Bell says. Based in Cary, North Carolina, FRG is not located in or near any of the major global financial centers, but it did have one valuable resource: a location adjacent to what’s known as the Research Triangle, anchored by Duke University, University of North Carolina, and North Carolina State University. 

“We could not grow this business without NEET,” Bell says. “That is the major benefit for us. We’ve found we cannot hire without this program—qualified folks just don’t exist at the numbers that we need to grow our business.” 

FRG runs two or three classes a year and they are small, with around three to five students. NEET has expanded to all of FRG’s locations, so over the past decade, it has graduated 42 students across the US, Canada and Malaysia, and it now partners with 20 universities. 

“The program is divided into two sections, one focused on technology and different tools students are going to need during implementation or model development, and the second is more about the domain, risk expertise—both in market and credit—things you can’t get in school, for the most part,” Bell says, adding that program participants earn certifications in those areas, such as FRMs or CFAs, depending on the candidate. 

NEET training lasts about one year, and kicks off with a 60-day “boot camp” designed to put students through their paces quickly, “mostly on the technical side, although there are some reading assignments on the domain,” Bell says. During boot camp, students take certification tests and once they pass, “they’re out of boot camp and into the advanced training piece,” which takes another three to four months. “This is where they take their basic technical skills that they’ve learned, we enhance it with some other courses and then really start injecting the domain—What is credit risk? Why is it important?—trying to apply what they’ve learned in the technology part to real business problems,” he says. 

Students are assigned mentors, and this phase is where it gets interesting, because candidates start work on current use-case problems. NEET also tosses in what Bell calls a little “light hazing.” 

“Last week, we dropped in on a Friday at 10 o’clock and said, ‘Okay, here’s an ask, and you have to deliver it in front of everyone at 2 o’clock.’ They have a short time window to research, typically without a lot of guidance—we may give them a little bit of data. Usually, we rip those out of things we’ve been asked to do by our clients, so they’re fairly well-grounded in real risk-related challenges, not pulled out of the sky,” Bell says. During this phase, students also work on soft skills such as technical writing and public speaking. 

To get to that point takes about six months, and candidates who make it are deemed ready to move onto project work alongside their mentors on smaller tasks. Bell says they try to keep it about 60% learning, 40% project work over the next few months, “until they ramp up to speed. The goal over that next six months is to have them ready to join a team full time.” Some classes stay internal at FRG while others are embedded directly with FRG clients. 

Inside the Vault

And of course, some end users run their own, similar programs, such as Mizuho Bank, which for the past three years, has partnered with universities to bring data science students into the bank to collaborate on data projects. “They get access to the datasets in a controlled way; we get access to new talent and new and innovative ideas,” says Gary Goldberg, chief data officer for Mizuho Securities EMEA, who adds that the bank is kicking off a new set of projects this year. 

An early, successful project was the development of an algorithm to identify data quality issues, with the aim to automate and reduce the time spent cleansing data. “We had a really interesting project done where a student actually started profiling our reference data and as a side effect, was able to self-identify the types of data through pairing and clustering.  After that, he would run the data quality algorithm against the historic datasets,” Goldberg says. 

Normally, data quality checks are done on the basis of mandatory attributes—for example, if in instrument requires a specific identifier that is missing, then that’s an error. “The limitation with rules-based data quality checks is you’re only ever going to capture anomalies based on rules that you have identified. What we’re doing with our analytics approach is different.  We use an algorithm to look at historical time series. It then builds up a pattern of normality so when a data point varies from that normality, it’s flagged as an exception.  We can rate the probability of an exception being valid based on the profiles of multiple attributes,” he says. “Those [data quality issues] get flagged up to the data maintenance team to be managed proactively.”

Mizuho’s university partnership program is attracting interest from around the bank. 

“That work has also led to interest from other parts of the company. People have heard about these data experiments and want to understand more about them. Data’s become cool,” Goldberg says. 

The projects vary, and he says the bank is cognizant of the balance between keeping proprietary information locked down while keeping the program engaging. 

“When we’re dealing with university students, we have to be careful about security,” Goldberg says. “We’re not going to give students access to confidential information, but we offer projects that deliver real value that gain the interest of students.”

Most of FRG’s clients are large financial institutions, energy companies and commodity firms. A NEET team is currently working to implement a mortgage model on the basis of a set of quantitative inputs. The client provides spreadsheets and some documentation, which the students scour to “figure out what they want to do and then implement that, which essentially means code it up in whatever language or technology the client is using, and then go through the process of testing that, vetting it and documenting it so it gets rolled out,” Bell says. 

Another recent project was to add some functionality to a user interface, which the client needed for regulatory reasons, and so students interviewed end users and assisted in writing the back end, but were not involved with actually building the functionality.  “That involves everything from figuring out how the database should work, any of the analytics involved and then of course, returning a report or some type of feedback to the user,” he says. 

NEET candidates have tackled client performance issues by taking a deep dive into a set of models to diagnose issues and return suggestions on how to improve. Students also helped a NEET client with a slow-running calculator by looking at the actual code and making a list of ideas for improvement, which were presented to the client for approval. “They’re currently working on executing and implementing those changes,” Bell says. 

Rhythm of Research

These partnerships are not limited to undergrads. Refinitiv Labs—the innovation division within Refinitiv—is partnering with universities to develop research that combines the data giant’s content and technology with the schools’ research interests. It’s been sponsoring a postdoc and his supervisor at the Alan Turing Institute for about a year as they look into trade anomalies and sentiment propagation. 

“We’ve discovered a problem area, we’ve funded them to do that research and we’ve supplied our data for them to do that research on real financial use-cases,” says Refinitiv Labs director Geoffrey Horrell, adding that it has a similar partnership with Imperial College London and it sponsors a senior lecturer at Bristol University. “We’re sponsoring that post and encouraging that department to use our data and giving them access to our information and to us, if their students or researchers want to understand more about how to apply their work to real financial industry problems.”

Horrell says that although the firm is open to expanding and increasing partnerships, “we’re very careful to select the right partner and the right project. It’s quite intensive. You really need to be invested and involved. We’re very open to new partnerships and new work, but it’s sort of on a case-by-case basis.” It’s taken Refinitiv about three years to establish its current slate of partnerships, because of “the timing and cycle of how university departments work. They have to publish research papers, they have to bring in Master’s students. They have their own rhythm. So we’ve had to line up the right kind of projects with the right kind of team. It’s taken us a couple of years to get a good working rhythm going.”

Nasdaq has another twist on an academic partnership, through a collaboration with MIT that provides the exchange with access to university research. “We also have been able to embed our projects in classes,” says Brad Peterson, CTO and CIO of Nasdaq, explaining that it runs internship classes twice a year; then if it hires those interns, they become mentors for academic partnership projects “and we run them like a product team.” 

The undergrads act as engineers, working on a real build, Peterson says. “And they’re dealing with MIT alums, usually, so they speak the same language and they’re living out of their apartment or dorm so they don’t have to move for the summer. I think we’ve found a good rhythm out there.” He says they’ve also expanded the partnership to Stanford.

Beyond Brainpower

Recruitment was the spark behind FRG’s program, but the partnership has resulted in other benefits, as well. “We seed graduates out into our own clients, so there’s this built-in FRG bias as our NEET folks move up the corporate ladder. We hope they fondly remember their days at FRG and then call us when they need help,” Bell says. “I have folks now who are fairly senior, leading teams, who went through our NEET program.” 

The positive impression is reflected internally, as Bell says NEET candidates who come to work for FRG tend to stick around. “When we started NEET, I figured we’d get three years, tops, out of them. A lot of them have had 10-year anniversaries already. There is a very low turnover rate for folks who have gone through the program.” 

One notable alum is the inaugural NEET graduate, who still works at FRG. “Our first NEET class was one person,” Bell says, adding that the candidate actually helped build the program material. 

“[NEET] gives us this nice pipeline of fresh ideas and thoughts constantly coming into the company,” he says, and the candidates are not limited to finance students, specifically mentioning a history major who graduated the program, was hired by a client and is still there. “It’s a testament to the program, when we can take someone who really had no financial background in school and was able to be very successful.”

Recruitment for NEET is deliberate and treated almost like a professional sports draft, Bell says, as professors recommend promising students—usually sophomores and juniors, but sometimes as young as high school—and then FRG maintains a database and tracks the potential candidates. “We also leverage the NEET kids as they come in,” Bell says. “It’s kind of this perpetual engine—as students come in, then they’re perfectly situated to reach back out to their professors and tell them about the program.” 

Refinitiv’s Horrell also sees building relationships with universities as a method to increase hiring power. “We want to bring talent from these top schools into the company. We want to explore new areas, push the boundaries, and discover new things. It helps us create new intellectual property. Creating defendable intellectual property is very valuable when you’re trying to grow in this knowledge business age,” he says. 

University partnerships bring marketing and promotional benefits, as well. Horrell says Refinitiv provides its research to other university partners. “If they produce research papers that use our content, we point our customers to it,” he says, adding that academics are often happy to speak to Refinitiv clients about their research. “Our customers like that independent perspective on some of these problems.”

NEET has been so successful that FRG is expanding the program to its sales force. It is also cost-effective, as clients who want a class embedded in their firms pay FRG to train and facilitate. “We make money on that, because over the second half, we start billing the client for the project time. That [second six months] margin is what pays for the first six months for us,” Bell says. 

Horrell points out that it’s not just the partners that gain from teaming up—projects that share data with academics have the power to elevate the entire industry. Currently, he says, a lot of the game-changing, published research on machine learning and artificial intelligence (AI) does not use financial data. “I think that if the financial markets want to really benefit from the financial research being done at these institutions, we need to open up our datasets to be used by academics. I really think that machine-learning models or AI models are only as good as the data they use. We really want to open up access to the data we have at Refinitiv to researchers so that the models that come out can work for the financial community.” 

Google, for example, could hypothetically publish a “fantastic” deep-learning model based on data from Gmail or its open news content, “but if you train that same model on financial news or financial research, it will be much more accurate when applied in our market,” Horrell says. 

He thinks it is a goal for the industry to make datasets available so research models are useful to financial services. “That’s where I see opportunities for us to do more,” he says.