UK watchdog to draw up guidance for AI in cross-border trades
FMSB aims to weed out misconduct in the global fixed income, FX and commodities industries
A UK body working to clean up financial markets will issue guidance on electronic trading for banks and other institutions, as technology such as artificial intelligence disrupts processes, its chairman said.
Financial institutions are adopting technologies to enhance productivity, better serve customers and reap higher returns.
In addition, they are facing scrutiny from regulators since the global financial crisis, and are required to report transactions and other cross-border business activity.
“Electronic markets are just as prone to abuse and manipulation as traditional markets are,” said Mark Yallop, chairman of the fixed income, currencies and commodities (FICC) Markets Standards Board (FMSB), based in the City of London.
The same people who previously would have thought of “spoofing” markets by manipulating them up or down to
illegally elicit the highest value from end-of-day trades, “now write computer programmes to do the same thing”, he said.
The FMSB was set up as part of the Bank of England and UK Treasury Department’s Fair and Effective Markets Review in 2015, which made 21 recommendations to weed out misconduct in FMCC markets, including lengthening jail sentences for those convicted of financial crimes.
The FMSB, whose membership comprises around 50 international banks, asset managers, corporations and brokers, has no enforcement powers but is tasked with producing professional guidance on how to handle grey areas such as information sharing and price setting.
Apart from his role at the two-year-old organisation, Mr Yallop is a member of the BoE’s Prudential Regulation Committee and was a UK group chief executive of Swiss bank UBS until 2014.
The Middle East’s FICC trading volumes are modest compared to those in the United States, Europe, South America and Asia. However, its financial markets are growing and becoming better regulated and local institutions are expected to join the FMSB in the coming years, Mr Yallop said.
He said that some emerging technologies, such as digital ledger blockchain, could increase transparency in financial services by accurately recording trades and clarifying price formation for the rest of the market.
However, the impact of technologies such as AI and machine learning – where computer systems learn from ongoing data input and start making their own decisions – is less clear.
“I defy anyone to tell us what the effect the widespread application of machine learning on financial markets will be, because it doesn’t exist at the moment and nobody knows what will happen if you have thousands of computers with different AI systems operating against each other,” he said.
“What happens if we reach a mini crisis in markets? What if there is an unexpected data release in the US, or a company in Japan goes bust, or suddenly the Chinese foreign exchange figures are revised to half what they were before – how do machines react to these sorts of stresses? The application of technology to financial markets is a probable good, but it carries scary implications we need to be prepared for.
“Who writes the code? A 16-year-old in his bedroom, or an army of ex-defence professionals?” Mr Yallop said.
Helping the industry understand those risks will be an important part of FMSB’s work over the next two to three years.
The organisation has issued guidance this year on governance of algorithmic trading platforms, and another paper on the management of trading “venues” (electronic stock exchanges) is under consultation and should be finalised in January.
But more work needs to be done, Mr Yallop said.
Financial markets will require governance on how AI trading systems are developed, how systems are supervised during trading hours, and what mechanisms are in place to halt them if required – the so-called “kill switch”, he said.
Updated: November 13, 2018 09:30 AM