Regulators must leverage private sector talent, technology and expertise to adequately oversee the deployment of machine learning in trading
By Belmont NewsBeat
To date, use of machine learning by the investment management industry has mostly been by compliance and back office functions such as anti-money laundering and know-your-customer checks.
The past six months, however, has seen a surge in the number of firms and FinTechs exploring the application of this new technology to front-end trading. While there are a number of players already using machine learning to execute trades, deployment across the industry has so far been somewhat limited.
This will soon change, predicts Mark Yallop, Chair of the UK’s Fixed Income, Currency and Commodities or FICC Markets Standards Board, or FMSB.
With the proliferation of new capabilities, he asserts, come new risks that regulators alone will be unable to oversee and control.
“The challenges posed by concentrated market structure, for example, can probably only be addressed through competition policy and law, informed by public policy considerations,” said Yallop at the 2019 Refinitiv Toronto Summit. “But other challenges — transparency, explainability, model risk management, governance, bias and correlation… cannot be solved by regulation alone.”
Machine learning, the FMSB Chair added, increases both the technical and the structural complexity of markets: It will be essential that private sector expertise, risk management and controls keep pace.
Yallop cites scarcity of skills, information asymmetry, cross-jurisdictional trading and the rapid pace of change as core issues for regulators but notes that they are areas of expertise the investment management industry has in abundance.
Transparency and explainability are particularly acute: Yallop explained the difficulty of identifying how trading decisions are made by machines, and how it is challenging to prevent in advance — or to correct afterwards — undesirable model outcomes.
He also warned that it is impossible to predict how a machine, trained on known historical data but making its own decisions will react when it is live in the market with a much larger dataset when it encounters events that haven’t been seen before.
Yallop cautioned of today’s skills gap: There is burgeoning need for expert programmers, data scientists and risk managers that can safely develop, test and implement machine learning in financial markets.
“These skills are in short supply in the private financial services sector and among central banks and market regulators. And they contribute to a quite significant knowledge gap among senior management, in the boardrooms of financial services firms and at policy makers about the hazards of AI [artificial intelligence]. This knowledge gap needs to be filled, soon.”
The FMSB Chair asserted his organisation must play a critical role in bring the public and private sectors together to better understand and regulate machine learning — he highlighted the board’s focus on this technology for the next few years.
“I believe we can and will make a very significant contribution to the safe deployment and realisation of the huge benefits that machine learning can, and should, deliver for the users of financial markets. And I hope that we can do this with many of you in the room here today,” Yallop concluded.
A full transcript of this speech can be downloaded at the FMSB’s website.