Data Platforms and Visualization, Finance
5 big data use cases in banking and financial services
For financial institutions mining of big data provides a huge opportunity to stand out from the competition. The data landscape for financial institutions is changing fast. It is not enough to leverage institutional data. This has to be augmented with open data like social to enhance decision making.
By using data science and machine learning to gather and analyze big data, financial institutions can reinvent their businesses. Financial Institutions are becoming aware of the potential of these technologies and are beginning to explore how data science and machine learning could enable them to streamline operations, improve product offerings, and enhance customer experiences.
Given our experience in Ideas2IT, implementing complex use cases for the financial industry, here we break down the top use cases of machine learning and data science in finance.
Catch stock market cheaters
Market surveillance depends on algorithms to identify patterns in trading data that might indicate manipulation and alert staff to investigate.
But the huge volumes of data can cause an enormous number of alerts, many of which are false alarms.
FINRA monitors roughly 50 billion events every day, including stock orders, modifications, cancellations, and trades. It looks for patterns in the events to uncover potential rule violations. But, most of the alerts received are false.
To tackle this issue, FINRA is developing a machine learning software that can look beyond the patterns and understand which situations truly deserve to be mentioned as red flags. In other words, the machine learning software will learn which trading patterns lead to legal charges, to classify the right ones.
FINRA is planning to test its machine learning software alongside its existing system to compare the results. It has also moved its market surveillance system to AWS cloud, giving it more computing power to analyze data quickly.
Detect phone fraudsters
Customers contact their financial institutions over the phone to check account balances, open new lines of credit, change account information. Mostly, a call centre agent facilitates the customer’s request. However, the agents have few ways to determine whether the person they are speaking to on the phone is the actual customer, and this poses a serious threat to that customer’s information.
In recent years, the scope of call center fraud has become truly staggering. In 2015, one in every 2,000 calls was fraudulent. In 2016, that number jumped to 1 in 937, an increase of 113%.
To solve this problem, Lloyds banking group partnered with Pindrop, an AI startup, to detect fraudulent phone calls. Pindrop can identify 147 features of a voice from a phone call or a Skype call which can help a person identify information like the caller’s location. The software will be integrated into Lloyd’s customer service offices. The banking agents will get an alert if the call is fraudulent so that they can pass the call to fraud specialists.
Lloyds banking group will introduce the software across the Lloyds Bank, Halifax and Bank of Scotland brands early next year.
Understand customers better
Today banks are using big data to create a 360-degree view of each customer based on how everyone individually uses mobile or online banking, branch banking or other channels.
A good example of this is Danske bank. The bank wanted to predict the needs of their customers and understand them on a more personal level. So, they created an in-house startup, advanced analytics, to transform business units using machine learning and AI.
The team analyzed large volumes of data to identify their customer’s preferred means of communication, such as phone, email, or social media. This valuable information has increased the hit rate of their marketing campaigns four times.
They also built a machine learning model to study the online behavior of their customers and discover situations where customers needed financial advice.
Streamline client payment processing
Reconciling payments is costly and time-consuming. Especially, when there are large quantities involved. Bank of America Merrill Lynch developed a new solution in August 2017 called Intelligent Receivables (IR) to help companies drastically improve their straight-through reconciliation (STR) of incoming payments.
Bank of America Merrill Lynch’s Intelligent Receivables, powered by High Radius’s leading-edge machine learning technology, will help their corporate clients to accelerate the adoption of electronic payments from their end-customers.
IR is a well-suited solution for firms that manage lots of payments where the remittance information is either missing or received separately from the payment.
Reduce financial crimes and parse commercial loan agreements
The Singapore based OCBC bank revealed plans to use artificial intelligence and machine learning as a part of its efforts to curb financial crimes. The bank plans to use these technologies to monitor anti-money laundering and to improve the accuracy in detecting suspicious transactions.
OCBC Bank along with ThetaRay, a fintech company, conducted a proof of concept (POC) at the starting of this year. Now, the bank plans to start an extended POC and pre-implementation phase. The algorithm will detect anomalies in transaction behavior by accessing different features such as products, customers, and risks. In the POC stage, the technology was deployed to analyze OCBC’s one-year transaction data, and it decreased the number of alerts, that did not needed further review, by 35%.
Interpreting legal and financial documents is a mind-numbing job for legal teams in financial institutions.
JP Morgan Chase & Co built a machine learning program called COIN(contract intelligence) to analyze financial deals. Before the project went live in June 2016, lawyers and legal teams spent 360,000 hours parsing commercial documents. Whereas, now, the software can review documents in seconds, and makes fewer errors.
The bank also plans to use COIN for other types of complex legal filings such as credit-default swaps and custody agreements.