Portfolio Management: The Holistic Data Lifecycle

Abstract

Machine learning provides many benefits to Portfolio Managers in analysing data and has the potential to provide much more. A concern with the approach to Machine Learning in Portfolio Management is that is caught between two domains: finance and information systems. In reality, to ensure its success, having these two separate and distinct domains are problematic. What is required is a holistic view, facilitating discussions, with data being the unifying concept and the one that is key to success. The data value map is a lens that allows all involved, in the use or adoption of Machine Learning in Portfolio Management, to form a shared understanding of the lifecycle of the data involved. Rather than being seen as a financial concept or a technical concept, this view of the data lifecycle provides a platform for all involved to determine what is required, and to identify and deal with any potential pitfalls along the way. A holistic view, and shared understanding, are required for the success of Machine Learning in Portfolio Management. Research on the intersection between Machine Learning and Portfolio Management is currently lacking. A focus on the different parts of the data lifecycle provides an opportunity for further research.