Everything about big data is big, including the challenges faced by institutional investors in managing it.
“Through a multi-modal empirical analysis of the data-management experiences of large institutional investors, we find that these entities are struggling to utilize data efficiently, and to consistently achieve desired levels of data quality.” That’s the opener to an August 2017 research paper* by Ashely Monk, of Stanford University’s Global Projects Center, and Daniel Nadler and Dave Rook, both of Kensho Technologies, a global data analytics and machine intelligence company.
Data is the beating heart of the markets, and the key to effectively harnessing big data is “turning it into information,” says Barb O’Malley, Senior Vice President, Northern Trust. “The key question facing institutions is: How do you do that? The short answer is that you have to understand the context of your data – where it comes from, the validity of it, what it all means.”
Matching talent to context
In order to help its clients understand that context, and reap the benefits of big data, Northern Trust is leading a trend by expanding the role data scientists play at the firm – a distinct role from what many investment professionals would think of as IT or tech support.
“Good data scientists are not traditional IT people – they’re going to be more statisticians, people from a quant background,” O’Malley says. “They analyze the data, determine the story it’s telling them, and then go back and refine their inquiry. It’s very much an iterative process of new queries to get a more meaningful answer and greater clarity.” This may all sound inefficient, but with modern open-source software, it’s quite the opposite, and data scientists can churn through data quickly while minimizing costs.
In layman’s terms, the process O’Malley describes is about filtering out the noise in the data. The talent to do so is highly sought in the investment world at the moment, and firms are looking beyond their own walls for it.
“There’s not a surplus of people with those skills. We’ve added a person who is a Ph.D. in mathematics and some others from the outside who have what I’d describe as actuarial skills. In some cases, we’ve also given opportunities to some of our IT professionals who demonstrated a natural tendency to dig deeper and continually ask new questions,” says O’Malley. “Math departments at universities are a good place to start looking for data science talent.”
The contributions of data scientists are most impactful when they are part of a holistic strategy to organize, validate, share, and distribute data under an umbrella of effective governance. This inevitably thrusts data scientists into relationships with seasoned investment professionals, and the interaction is often a new experience for everyone involved. It has also produced intriguing outcomes.
“You get some pretty powerful insights by bringing those two mindsets together,” O’Malley says. “When someone has spent a fair amount of time in a particular part of the business and they come to us with a problem, they often have a preconceived notion of what the answer should be – perhaps because they are too close to the subject. The data scientist doesn’t have that bias. They come in and say, ‘I'm going to go crunch things through and let the data tell us the answer.’”
Collaboration of this type sometimes results in proactive solutions for clients and makes markets more efficient. For example, when Northern Trust saw potential value to its clients if it could improve the efficiency and transparency of foreign exchange (FX) transactions, data analysis identified an opportunity and showed that the capital investment was worthwhile. The result was the creation of an algorithmic trading service that delivers significant benefits to clients when it comes to FX activity.
The big data horizon
Given the sheer volume of data handled by businesses around the world, the risk of inappropriate disclosure – either mistakenly or maliciously – has pushed the subjects of transparency and governance to the forefront of global regulatory conversations. The EU, for its part, recently implemented the General Data Protection Regulation (GDPR), a sweeping act that required businesses, including those in the investment world, to comply with a host of requirements, such as encrypting “data at rest,” i.e. data sitting idle is encrypted so that even if it’s improperly accessed it cannot be interpreted.
As data becomes more sophisticated and artificial intelligence (AI) and neural networks play a greater role in analysis regulators are, understandably, going to have more to say on these issues. The financial sector may be earlier on the adoption curve than, say, the retail sector when it comes to gathering and analyzing behavioral data, but it is starting to make serious inroads. This will only add to the challenge of big data analysis – more data means more to analyze – and it’s a given that AI will grow in importance as companies look for ways to meet this challenge. For machine learning to be optimally deployed, the machines must continually learn, and that will likely create new job descriptions – to say nothing of new ethical challenges.
“Pretty much every firm I talk to understands the difference between data that is ready for consumption and data that is still raw and should be used cautiously,” says O’Malley. “That’s where the ethics comes in – to make sure that you can fairly articulate what controls you have to make sure the data is being utilized appropriately. We’ll start to see data ethics taught more and more at universities, and it will be addressed by regulators, too. Because everybody is looking at AI and big data and saying, ‘Isn’t this cool? Isn’t this wonderful?’ And it is, but we always have to be able to answer the question: ‘How is my data being used?’”
*Data Management in Institutional Investing: A New Budgetary Approach