The banking sector has been leveraging data science capabilities in order to accelerate operations and increase flexibility. As an industry with highly sensitive data at its disposal, data science has also played a role in strengthening security, particularly through enhanced identity management. In this article, we explore the biggest data science trends that have emerged within the banking sector.
Automation, cloud and ESG
- Automation: “Many key insights and correlations across finance can now be automated using machine learning. This enables investment professionals to analyse much larger data sets (from hundreds of columns to millions of columns) and new sources of alternative data (e.g. social media, satellite imagery, credit card spending data, weather data) about economic activity, enabling them to more quickly make informed investment decisions and develop new investment strategies.”
- Cloud computing: “The cloud gives banks and other investors the ability to immediately access and store data and compute resources, so they easily scale their operations on-demand. Bloomberg’s ability to seamlessly deliver our market data feed (B-PIPE) quickly and reliably around the globe to our clients’ internal and third-party applications via the cloud provides them with potential cost savings and performance enhancements, while enabling them to generate insights and analytics in a more cost-effective manner.”
- ESG data: “The increasing availability of Environmental, Social and Governance (ESG) data is enabling investors to develop strategies that focus on the societal impact of companies. More broadly, the ability to build thematic investing strategies like this are enabled by AI, which helps investors better analyse and understand which companies are impacted.”
Use of alternative data
Alternative data, which has been utilised for the benefit of investment processes within hedge funds and investment banks, is now making waves in banking. Hosted beyond the parameters of traditional data sources such as financial statements and broker forecasts, alternative data provides banks with deeper, more timely insights for aiding business decisions.
“We have seen a surge in inquiries from companies within the financial sphere,” explained Julius Cerniauskas, CEO of Oxylabs. “According to our calculations, the number of inquiries has increased almost threefold over the past year, indicating a quickly rising interest in alternative data acquisition.
“The banking industry is increasingly venturing beyond traditional financial statements, company filings and management information. One of the biggest trends in the banking industry is the rise of alternative data use in investment decision making and ESG analysis.
“While hedge funds and investment banks have been using alternative data for years, it is now spreading further across the industry and the scale of such data use is increasing. Alternative data can provide signs of market movements through, for example, the company’s hiring pace, hiring of high-level executives or customer sentiment.”
Data ethics: how to work out what’s right for your business
Traceable timing solutions
Another rising data science trend within banking is the use of traceable timing solutions. With timestamping regulations in financial services getting stricter, and data scientists looking to maintain integrity of assets, these solutions look to improve the accuracy of time sources.
Richard Hoptroff, CTO and founder of Hoptroff, said: “Network derived and precise traceable timing solutions are an innovative aide which can be used to bolster data infrastructure of banks. They are a growing alternative to traditional, satellite dependant means of achieving time. Network derived time can be used to optimise trade lifecycle management, improve transaction reporting and inform strategic decisions. This enables the verification of transactions to become more efficient and reliable and also opens up the possibility to identify significant cost savings.
“The implications of the Fourth Industrial Revolution mean that traceable timing solutions are becoming increasingly relevant outside of financial services – as posited by Brad Casemore, the vice-president of Datacenter Networks at IDC, ‘Time and time services are more ubiquitous and more valuable today than many business leaders realise’.”
While the evolution of data science, including artificial intelligence (AI) and machine learning (ML), is showing no signs of slowing down, and frequently bringing new business use cases, the planned return on investment (ROI) still needs to be clearly explained to the board.
“Gartner advises that 80% of models do not make it into production and there is a lot of waste, duplication and sheer resource intensive.
“As a result, ML return on investment is the hot topic and needs to be the primary focus as boards are demanding more business value from the data science teams. Gartner predicts that by 2022, 90% of banks will explicitly mention AI as a core analytical competency and investment will grow. Boards will look for a return on this investment.
“One of the drivers behind the issue to operationalise models is access to good quality data, so the trend is to move away from model-centric AI towards data centric AI. This is supported by Andrew Ng, and is core to Teradata’s Analytics 123 proposition and our Enterprise Feature Store (EFS).”
Hot topics and emerging trends in data science
The two sides of data analytics
Finally, there is the general use of data analytics in the banking sector to consider — increasingly utilised to bolster identity and risk management, it is by no means a silver bullet, and can even harm the user experience if not properly monitored by data scientists.
“The use of data analytics for decision-making, fraud analysis and risk detection has been game changing but it can also be something of a double edged sword,” said Franki Hackett, head of audit and ethics at Engine B.
“For consumers, data makes the process of accessing financial products much faster (because banks can now swiftly make data-driven decisions), and means a bank can offer its customers appropriate banking services and products.
“There are risks with this, however, as we know that when banks algorithmically make decisions it can make them more prone to biased outcomes. It can also mean that banks can become unwilling to actually speak to their customers in order to glean more information, or to provide support and advice.
“This can make people who have characteristics of vulnerability less able to access banking services and we need to ensure that the human element is retained along with the use of data.”