The views and opinions expressed in this article are those of the thought leaders as individuals, and are not attributed to CeFPro or any particular organization.
By Julian Horky, Head of Risk Controlling, Berenberg Capital Markets
What do you see as some of the key challenges within models and algorithms in a broker dealer?
Automation and artificial intelligence (AI) are unquestionably among the most exciting topics in the financial industry. Whether you talk to the buy-side, sell-side, consultants, regulators, or even academics, it’s the topic of the moment in the industry. You hear especially often how explainability of AI models is a central challenge. This challenge applies to users, developers, validators, and auditors just alike. Therefore, industry and regulators are working together to create a framework that allows rapid adoption without lowering the bar regarding model safety.
However, there are more subtle issues that are equally important even if they do not receive as much attention. My statistics professor was always talking about the fundamental problem of spurious correlation in large high-dimensional data sets. Also, false positives and negatives pose a problem – not only in fraud detection but also in medicine and other fields. The essence of these topics has made it out of academia and is now widely understood by the professional community.
