Download the authoritative guide: Cloud Computing 2018: Using the Cloud to Transform Your Business
Big Data is playing a growing role in financial services in several ways. Specifically, predictive analytics and real-time decision making is becoming more of a reality to financial advisors and their clients, even in a sector where past performance is no indicator of future behavior.
Making financial decisions, whether it’s to purchase a stock or give a loan, requires data points, and the more data available to the client or bank, the more accurate of a decision that can be made. Big Data has opened the spigot as it were, bringing in a flood of information.
As always, data is useless in and of itself. It’s how you use it. And financial services firms are using Big Data in some interesting ways.
Big Data for Financial Analytics
Ben Musgrave, principal consultant with Synechron, a business consulting firm that advises financial firms, said he knows of one bank that gets a billion data points a day from its infrastructure and at the moment they are not making the most of that.
But some are. “There are banks on the investment side more proactively reacting to clients’ wants and needs where there will be hundreds of thousands of orders from clients, and they are mining orders for patterns. They look at what industries are they investing in, what geographies, and so on, and comparing that to what’s on the news. So they can quickly map back [items of interest] to the client and beat the competition,” he said.
He also knows of a private bank where they are mapping the internal repository of the holdings clients have and taking in a real-time news feed and an inventory of trading ideas and then mapping those together. So when a private banker comes in first thing in the morning, instead of having to catch up on the news, the news is mapped to his client holdings and he gets an alert saying to talk to a customer because these news stories could affect their investments.
Netvibes CEO Freddy Mini said Big Data is the end of emotional decisions. “It’s the same difference between instinct and insight. Instinct was you rely on people with a large experience and gut feel and a guru you listen to and you believe the guru has the right way to go. And everything is based on emotion, and in my opinion that’s a recipe for failure,” he said.
Netvibes works on finding insights that will put you ahead of everybody else based on data and facts and algorithms, he said. “So it’s the exact opposite of emotion and bias,” said Mini.
One example: he noticed at the Consumer Electronics Show in January that there was a lot of talk about products similar to Google Glass, Google’s attempt at an augmented reality headset. This gave Netvibes the sense that Glass would come back. Just a few weeks ago, Google announced version 2.0 of Google Glass as it jumped back in the fray for augmented reality headwear.
Big Data Benefits
The flood of data is forcing changes at banks in a number of ways. Kaushik Deka, CTO of Novantas, which specializes in banking industry analytics, said the number of data sources has exploded, both in structured and unstructured data. That has put a little stress on the models that requires a new way of thinking.
“Now they have to think how they deliver insight back to banks and customers. It’s forcing them to think in new ways and how to harmonize different data sources to utilize new analytics,” he said.
The focus is more customer-centric, whereas before it was more product-centic. “It’s all about customer behavior analytics and data related to the customer. Banks have moved from product pricing to customer-centric pricing,” he said.
The other change is around the delivery of analytics. It used to be a batch process would be run on models on a weekly or monthly basis and decisions were made weekly and monthly. Now the real time nature of positioning has come into play.
“Models are run more frequently and decisions are made more frequently. There is still an element of human judgement made in decision making. We did not take human element out of the equation. With expertise, decision making has changed from batch to real time,” said Deka.
Musgrave said the explosion in data is giving clients a much more personal experience by saying, in effect, “this is a price I’ve come up with for you,” as opposed to giving one price to all clients. By mining unstructured data like emails, they can compare email sentiment to things like when you made an investment and compare that against news events. If they all map together, the bank might suggest a transaction based on the user’s sentiment, history, and corresponding news.
“It’s very specific to you, but if I extrapolate that over my client base, I can do client clustering so I can say you are similar to someone else, so if I get something out of you I can try the other guy. So you can try that next action, this is the next thing you should do, this is the optimal solution,” said Musgrave.
Another benefit of the data intake is improved security by listening to places like the dark web and other places where illicit activity occurs. “We listen to what people are saying. If you have people saying here is the fake account, here’s how to crack the systems, how break into accounts, we can compare what alerts and reports for banks, and have a response,” said Mini.
Considerations for Big Data and Finance
Big Data is helping security be more accurate and deal with false positives. With tens of thousands of transactions going through every day, you can’t stop them all for an alert, but you can do analysis and allow it to go through, said Musgrave.
“A bank that I spoke to said they have thousands of alerts every day and 99% are false positives. With thousands of them every day, you have a matter of seconds to decide if it should go through. They get false positives because most of the screening is done on very simple rules,” he said.
With the technology available from Big Data you can do far more analysis and faster. Now you can apply a lot more context to keywords as opposed to check for a name or country. More data is available and you can do it faster and that speed is really key, he said.
The issue of privacy is one the industry has to deal with, because Big Data’s natural openness clashes with regulatory requirements for privacy. Financial services is one of the most heavily regulated industries out there, and machine learning can’t be used in thins like Comprehensive Capital Analysis and Review (CCAR) models, said Deka.
Regulation mandates say any model they do in Big Data to get to the final insight has to be able to trace back the translation logic to the beginning so they understand each step of the transformation. Because of this, they can’t plug in a neural network for all use cases because they don’t document each step along the way.