Machine Learning in Finance: A Novel Scientific Approach, or Just the Anecdotic Success of a Black Box?
Dr. Bastian Bergmann, Executive Director, ETH FinsureTech Hub
The fascinating successes of machine learning (ML) in language processing, image recognition, or multiplayer gaming have triggered many fantasies of applying these technologies in other fields, including in aspects of banking and finance. Often, ML methods give the impression of being black boxes, devoid of domain-specific knowledge but delivering almost remarkable performance. But in banking and finance, the unexplained, impressive functionality of ML applications is not enough—basic questions of algorithm explainability and rationalization must also be explored.
To understand the recent, promising advances in ML, a profound understanding of the underlying technologies is also required. One has, not only from an application perspective but also from a more scientific standpoint, to critically review these novel technologies. And in the process, it may prove necessary to transform our understanding of finance as a social science.
Objectives
In this SFI Master Class we will cover some of the basic concepts of ML and its most common tools and programming techniques, in view of its latest applications in banking and finance. By doing so, we will be looking into the basic building blocks of ML, including objective function, approximations, regularizations, and reinforcement learning. We will critically examine these according to scientific standards common in finance. The area of conflict between explainability, on the one hand, and performance, on the other, will be discussed in important ML case studies, including of deep hedging and deep simulation.
Target Audience
The Master Class is aimed at everyone who wants a deeper knowledge of machine learning and its potential for the modern financial industry. An interest in the conceptual foundations and in the philosophy of ML or AI is a welcome attribute for participants. Participants from a broad range of backgrounds—including economics, finance, or quantitative finance—are welcome. No programming skills are required, but one should not be afraid of discussing code.
SAQ Recertification
This SFI Master Class has been only recently developed. Its accreditation as an SAQ CWMA recertification measure (4 credits) is in progress.