Commentary: Growth of data science in financial services

Money managers are addicted to data. Driven by a need to increase efficiency and effectiveness in their ability to make predictions, money managers have talked about, considered and held data science in thrall for much longer than other areas of financial services — and in much more depth. For decades, analysts and managers have pored over numbers trying to ascribe meaning. Some used "technical methods," others "fundamentals," but the principles were the same — look for patterns, signals and indicators that help predict the future and generate value.

Regardless of this long and important history, some fundamental changes have happened in the last two decades that mean we are only now seeing a new phase in the way data are analyzed. To understand why that is, we need to briefly consider the technologies that underpin the way we work with data and how these have changed over time.

By in the early 2000s, data systems were widespread and commonplace in financial services — often customized or developed in-house — powered by large and expensive development teams each building their own platforms from the ground up. Physical infrastructure, operating systems, databases, analytics tools, reporting platforms — were all customized and built-to-order. While competitive advantage was achievable, this approach was not conducive to generating either scale or interoperability between systems. Without these two things there was no possibility of assembling the pools of big data needed to make viable the techniques we are seeing now.

In 2021, some of the core technologies that organizations use to produce vast volumes of data are now reaching venerable status, and we are living in the big data era. For example, this year marks the 15-year anniversary of cloud computing availability at scale and 16 years of mainstream Python adoption. These tools might be obscure to those outside of the technology industry, but these are major examples of some key technological pillars that have allowed companies to produce and manage huge volumes of data about their clients, suppliers, organizations and their products over the last decade and more. Critically, they are scalable, freely available and interoperable. Big data has become easy, reliable and available to all. This development has led to the viability of data-analysis techniques that are fundamentally different from historical ones — namely those based on machine learning.

It is hard to understate the importance of these new techniques and the elevated insight they provide for the financial-services industry. These techniques do not only provide direct answers to problems, but can act as accelerators and recommendation tools to support human decisions made by investment managers; they can provide insight and economical analysis that open up mass-market decision-making at all sizes of investment portfolio. The constraint of these tools, however, is that their effectiveness and their transformational ability is directly correlated with the amount of data available for them to work on. It is only now that we have enough data available, in a scalable way, that they can be effective.

Today, tools to query, interpret and make sense of this sea of data are much newer than those to generate it and are just reaching maturity — the most advanced machine learning developer functionalities (tools like TensorFlow, the industry-standard machine learning tool set) have only reached mainstream adoption in the last few years. The very recent and rapid availability of these tools has created opportunities and challenges — business expectations for what can be achieved by data interpretation have increased thanks to the progress of leading, innovative technology start-ups. Yet, the relative youth of the tools used to undertake this interpretation at scale means that the skills to use them are not yet fully embedded in the workforce.

Today, financial services organizations have high expectations of what can be achieved with machine learning and other data-driven technologies. A common ambition — and belief — is that these tools will support them changing their core business models to financial technology from banking services.

One of the largest critical challenges to address is the availability of skills. For example, Goldman Sachs Group Inc. recently announced the opening of its biggest office in the U.K. outside London, creating hundreds of technology jobs in Birmingham, England. In recent years, Birmingham also became the home for 1,000 employees from Deutsche Bank AG, mainly in back-office and technology roles. Goldman is hiring software engineers, data analysts and data scientists to work on new ways of delivering financial services at the Birmingham base, due to open by the end of the year. Most critically, these roles will not be working on customized services but common platforms, ideas and tool sets.

Some of the skills required to deliver on the data-science paradigm are not yet fully apparent or widespread — a widely understood example of this skills gap will be the ongoing need to understand the ethical use of data. So, compliance departments of the future may well be staffed by data ethicists who monitor the organizations' data scientists to ensure appropriate behavior, particularly as legal frameworks struggle to keep up with the pace of change.

But even when it comes to immediately required skills, recent research has shown a chronic shortage of data talent. In the U.K., for example, it is costing organizations more than £2 billion ($2.77 billion) a year, according to Nesta, a non-profit organization that fosters innovation. The government's National Data Strategy is seeking to establish the U.K. as a world-leading data economy and highlights the need for firms to develop the core skills required to analyze data sets and bridge the data-science skills gap, and this initiative has relevance globally.

It is now the skills gaps, not the technology gap, that will challenge organizations in the upcoming race to leverage data in the investment management organization.

Kevin Poulton managing director of product strategy at Fitch Learning, a professional training and development firm, which is part of Fitch Group. He is based in London. This content represents the views of the author. It was submitted and edited under Pensions & Investments guidelines but is not a product of P&I's editorial team.