Swissquote Conference 2018 on Machine Learning in Finance

Machine learning has made tremendous progress over the past decade. This is owing to the availability of massive data, rapid growth in computing and storage power, and novel techniques such as deep learning. Machine learning has also come to play a prominent role in modern finance. Applications range from textual analysis of business reports to deep learning for risk, investment, and operations management. But not every standard machine learning solution is appropriate for financial data, which exhibits noise and behavioral elements, as well as non-stationarity.

The conference featured seven presentations by leading experts in finance and machine learning from academia and industry. Rama Cont from the University of Oxford presented a deep learning approach applied to a high-frequency database containing billions of electronic market quotes and transactions for US equities. He uncovered nonparametric evidence for the existence of a universal and stationary price formation mechanism relating the dynamics of supply and demand for a stock to future variations in its market price. Artur Sepp from Quantica Capital AG illustrated the strengths and limitations of machine learning in quantitative strategies from a real-life hedge fund perspective. He demonstrated that a key problem in volatility based trading strategies has been and still is the selection of best models. Hans Buehler from J.P. Morgan presented a framework for managing derivatives under market frictions using reinforcement learning techniques and model based simulations. He demonstrated the power of their approach by a use case for hedging a book of exotics under market frictions. Kay Giesecke from Stanford University presented new statistical tests to assess the significance of the input variables of a neural network towards explainable artificial intelligence. He also showed experiments using real estate transactions data that illustrate their good properties. Gerard Hoberg from the University of Southern California presented a collection of research projects that use textual analysis in finance to construct models of industry structure and innovation. David Shrier from Distilled Analytics, MIT, and University of Oxford provided a thought-provoking discussion of ethical aspects of artificial intelligence. The discussion was centered on the question of how to ensure the ethical application of machine learning to financial services. Isabelle Flückiger from Accenture gave firsthand insights in day to day applications of Natural Language Processing in the financial services industry. Examples included research and due diligence for M&A in investment banking, the many investor reports, market information for trading as well as the analysts, and the processing of contracts and search for clauses and regulations.

The presentations were complemented by four invited FinTech companies (EdgeLab, EUROFIDAI BEDOFIH, Predictive Layer, and Swissquote) who showcased their products and services during the coffee and lunch breaks.

The conference attracted 250 participants, 60 percent of which came from the financial industry and 40 percent from academic institutions. It was jointly organized by the Swissquote Chair in Quantitative Finance, visiting professor Alexander Lipton, and the Swiss Finance Institute at EPFL. Sponsoring by Swissquote and the Swiss Finance Institute is gratefully acknowledged.