It’s great to see regulators keeping up with this rapidly changing landscape. This week’s OSC report gives a great overview of the current state but we do want to call out a misconception that has made its way into the report. Footnote 11 states: “… achieving the same level of accuracy as conventional simulation-based pricing models is not feasible when using AI.” This is simply incorrect.
The statement reveals some confusion between theory and practice. It is theoretically correct that traditional simulation methods eventually converge to the exact solution. In practice, banks don’t have infinitely long to calculate prices and have to settle for some amount of numerical error. Since prices are calculated so frequently, and the associated costs are so high, banks have learned to live with a significant amount of numerical error.
The error in the AI approach can be made smaller than this, resulting in the situation where the AI approach is both faster AND more accurate. If you want proof of this, check out what we are doing at Riskfuel. We’re the first to use AI to improve the efficiency and accuracy of market risk management.