RiskMinds Americas brought together 200+ CROs, regulators, academics and risk experts to discuss the critical pain points facing the industry now. Here, Peter Zeitsch, Solution Architecht at Calypso Technology, discusses solving BCBS-185 with Machine Learning.
The Basel directive, BCBS-185, outlines sound practices for back-testing counterparty risk models to ensure they are effective. Here, Peter argued that it essentially reduces to one tricky requirement: capturing the ‘black-swans’ or jumps. Taking fixed income as an example, Peter showed that Machine Learning can, not only identify the jumps, but recognise clustering within those jumps that correspond to actual events in the market, leading to new insights into how to build risk models that incorporate the features of the Machine Learning. It also calls into question some of the most basic assumptions in finance theory, which Peter will also explored.
