“The application of advanced analytics to specific business problems has begun to deliver value for traditional asset managers — not by replacing humans but by enabling them to make better decisions quickly and consistently.” This quote comes from a McKinsey report on data and analytics in asset management, and subtly nods to the gradual rise in data and analytics adoption across the asset management value chain.
However, the bulk of this steady adoption comes from the pioneers, the large buy-side firms who have taken risks to get ahead of their smaller and mid-size peers. Some large players, though, are lagging behind as they are much heavier organizations and much more fragmented across roles, with quants and data scientists solely focusing on specific projects.
The asset management sector as a whole is only now truly gaining momentum when it comes to integrating a more data-driven approach to its investment processes. From investment research and decision making to trade execution and portfolio management, the transformative capabilities data science and machine learning offer are significant — not to mention there are massive efficiency benefits to be reaped on the operational and control front.
In order to differentiate themselves and stay relevant among the sea of competition, asset management firms need to smoothly transition from simply discussing the theoretical benefits of data science, machine learning, and AI and actually put it into practice in order to see business value.
Why Asset Management Firms Need Data Science and Machine Learning Platforms
At their most fundamental level, data science and machine learning platforms are tools that enable Enterprise AI by equipping people (from diverse backgrounds and skill sets) within the organization to:
- Use data to produce predictive analytics (or machine learning) solutions.
- Scale rapidly by providing reusability, reproducibility, and traceability both across teams and various projects.
- Access all data and data projects from one central location (mitigating the risk of data silos or unintentional exclusivity, for example).
Further, a data science team is never going to look the same from one organization to the next. For asset management firms, for example, teams may work on developing tools and generating insights to help support factor-based investment strategies for clients and the firm, ranging from macroeconomic forecasts to sector and individual stock analysis to societal trends.
The team composition may look something like this: a rapidly growing team of five to 10 data scientists with some portfolio management and engineering experience, embedded into the front office among quants, traders, and portfolio managers, with plans to scale to 20-30 data scientists in cross-asset pod teams with dedicated DevOps specialists.
How, though, are they expected to communicate and share information from their data projects with the portfolio managers, quants, and traders? The answer lies in an end-to-end data science platform, which promotes the scalability, flexibility, and control needed for cross-functional teams within an asset management firm. This way, data scientists and portfolio managers can build mutual trust and jointly work on signals and models. The collaborative-driven nature of Dataiku's platform, for example, is a key solution to this concern.
Successfully adopting and implementing an AI — and a platform to efficiently usher in these efforts (while bringing significant time savings) — involves thoughtful, enterprise-wide and operational risk considerations for asset management firms. Data science leaders in this space should take the time to understand and identify their challenges and, conversely, the impact that is more frequently being observed when teams integrate data science and machine learning into the investment process.