Posted by Emily Leinbach on April 5th, 2018
Originally written on LinkedIn by Jim Urquhart, Managing Director, FinTech at The Bowdoin Group.
Last week, The Bowdoin Group, in partnership with venture capital firm Hyperplane, hosted a lively panel discussion about “Transforming Data from a Liability to an Asset” in Cambridge, Massachusetts. The panel brought together leading experts with diverse backgrounds in both finance and technology: Subbiah Gopalraman, Founder of IT Strategies Group, Jonathan Owen, Data Scientist and Quantitative Developer from Columbia Threadneedle Investments, and Slater Victoroff, CTO at indico.
Hyperplane’s Managing Partner, Jack Klinck, moderated the event, which played out with audience participation from an interesting mix of professionals from both large financial firms and smaller startups focused on applied data science. They had a common goal: Figure out how to harness the power of unstructured data and apply it to the trading lifecycle.
The group talked about challenges, trends, and best practices, and it generated a number of key insights. Here are my top seven:
1. Mining unstructured data requires new ways of thinking. Companies across industries have worked with structured data for quite some time, and for the most part, they’ve mastered its use – finding it relatively easy to tap into the information it represents. Turning unstructured data into usable, impactful information, however, is a new frontier. Firms struggle to manage, analyze, and ultimately use this type of data to improve processes and performance. Slater Victoroff commented that the models developed for structured data completely fall apart when trying to apply them to unstructured data. Maximizing the value of unstructured data requires new perspectives and innovative techniques, like natural language processing, artificial intelligence, and data science expertise.
2. It’s important to understand the problem you’re trying to solve. The panel emphasized the need to set finite goals when approaching any strategic data initiative. What do you want to get out of it? What outcome are you studying? There has to be clarity of the problem. At the end of the day, it’s about understanding how to cull through massive amounts of data to translate it into usable insight – about tying qualitative information to quantitative data. This is a lofty goal, and you must be clear about how you will measure success.
3. You’ll face old and new challenges. Many of the challenges that plague the management and use of structured data also apply to unstructured data. One attendee brought up the omnipresent issue of security. Panelists agreed that “mixing and sharing” data is unacceptable. Even where data is stored (for example, in-house versus the cloud) brings up critical security concerns. A newer challenge most face with unstructured data is that of ROI clarity. It’s still unclear what impact these new sources of information can have – or are having – on performance. The data is often difficult to use and the information it represents commonly mistrusted.
4. Technology can help, but presents its own challenges. There is a healthy skepticism about technology in this area, and a strong best practice is to start with a pilot or proof of concept before investing heavily in a technology-driven solution. It’s also important to consider how technology can be applied across the organization and how you can measure its ultimate impact. One of the biggest conundrums Financial Services firms face is that of “buy versus build,” with many looking at hybrid models. They’re also cultivating in-house expertise to understand up-and-coming technology solutions.
5. Streamlined processes are essential to realizing value. To get the most out of unstructured data, it must flow effortlessly into and through the Portfolio or Fund Manager’s workflow. It must be highly accessible and easy to use. As with any change that impacts the fundamental way a person does their work, expect to encounter some resistance.
6. Your workforce will evolve. When it comes to talent in both technical and financial companies, the group largely agreed that it’s not necessary for technologists or data scientists to be financial gurus (although that is a definite bonus). In fact, the panelists agreed that people without financial expertise often bring a valuable perspective, because they aren’t constrained by traditional biases. Firms are seeing an interestingly symbiotic relationship emerging between IT and business pros: Lines are blurring between roles, and we see data experts working hand-in-hand with Portfolio Managers.
7. What’s coming next? The panel was asked about the hottest trends coming in the next nine to twelve months. Here’s what they had to say: There’s hope that there will be more data science services available, which would give firms the ability to try out solutions more flexibly and take advantage of “on-demand” capabilities to save modeling time. Information overload is a problem, making automation another important trend. Financial services firms need automation to summarize content, marry it with sell-side research, and provide relevant social understanding. Finally, in the near future, we will stop talking about unstructured data as a data science problem – instead recognizing it as the business problem that it is.
How is your company approaching the management and use of unstructured data? What are you seeing in the field?