Twenty years ago, regulation was aimed at investment bankers; now engineers are the target. “The idea is to bring financial-engineering tools to financial regulation in order to improve the financial system as a whole,” says Mathieu Rosenbaum, chair of Analytics and Models for Regulation at École Polytechnique’s Centre for Applied Mathematics.
According to Rosenbaum, one problem machine learning can help solve is setting “tick sizes” – the smallest price increment that trading instruments can move. The Euro FX futures market, for example, has a tick size of 0.00005, which means prices may move only in increments of that magnitude.
Getting the tick size right is important. When set too small, the result is very unstable order books, lot of high frequency trading activity, uninformative prices and bigger differences between bid and ask price. Set them too large and you get price sloppiness, large queues in order books and trades are made less frequently.
In Europe, tick sizes are set according to the profile of each security (number of transactions per day) and their price. But this is an exact approach. “We want an automatic technological solution,” says Rosembaum. “This would represent a major advance in financial regulation.”
Professor for Analytics and Models for Regulation at Centre for Applied Mathematics, École Polytechnique Paris