AI model ‘better’ at pricing currencies than humans

LONDON – ING Groep NV is starting to use artificial intelligence (AI) to price currencies, replacing a job traditionally performed manually by the bank’s traders.
The Dutch lender’s new AI model employs “reinforcement learning” — a technique that mimics the trial-and-error process humans use — to make pricing decisions to keep up with market volatility, said Simon Bevan, its global head of electronic trading, in an interview. That was previously a crucial yet time-consuming task for its trading team in London.
“It’s a full-time job monitoring the market, adjusting spreads and managing the risk, so it’s freed up basically a whole person,” Bevan said. “This model completely takes care of that and has performed way beyond our expectations, it has definitely outperformed a human.”
The move is the latest in a race by banks to develop the most cutting-edge technology to reduce expenses and be competitive in the US$7.5 trillion-a-day global currency market. The next frontier is developing AI, which promises to increase efficiency and reduce the number of traders banks need to employ.
ING hired James Robinson, with a PhD in machine learning, from UBS Group AG’s electronic FX trading team five months ago to develop the technology. The overall build out took three months before six weeks of testing. Robinson is now working on creating other AI programs.
“It makes sense to take what we’ve done and see how we can use it in different asset classes,” Bevan said. “Working on more AI models will be a big focus for us going into next year.” 
Red button
These industry developments mean the day-to-day role of traders has already changed. Dealers at a recent conference were sceptical that AI will fully eradicate the need for human oversight and raised questions around the growing reliance on computer models, such as who will ultimately be responsible if things go wrong.
Bevan said it will still be the duty of the trading team to monitor the new program and stop it if it malfunctions. While ING has a strict model governance approval process, this went surprising smoothly given the controls around it, he added.
“We will always still have human oversight and a big red button we can hit,” he said. “The whole process is very deterministic, as long as you can explain it and document it, people are very comfortable with it.”
ING is rearranging its trading team to leverage the technology the firm has developed for its FX division to other asset classes.
Bevan became global head of electronic trading this month, overseeing equities, rates, credit, commodities, financial securities and FX, while Kimiya Minoukadeh was appointed global head of quant trading. Previously the pair oversaw FX.
“The speed of change within the FX landscape makes accurately measuring and reacting to these changes with traditional algorithmic models challenging,” Minoukadeh said. This sort of new AI-based algorithm “has vast applications across financial markets,” she said.