AI is All over the Forex Industry. Will it Push You out the Door?

Artificial intelligence is exerting a growing influence on FX trading: from its use in data analytics to sift through market information, to spot patterns and offer guidance on potential market movement, to the application of algorithms incorporating natural language processing, to analyse sentiment from the mountains of data generated by market participants and observers.

Machines Not Ready to Take Over Just Yet

The technology is additonally playing a part in liquidity management by identifying potential market flows, enabling traders to execute at the most advantageous time, and giving them insights into currencies that are being heavily traded. This information plays a part in reducing trading risk by highlighting periods of likely heightened volatility.

According to Bart Joris, the Head of FX sell-side trading at the financial market data and infrastructure provider Refinitiv, FX teams perform better when data and AI techniques are coupled with human insight. New skill sets are becoming increasingly important so that individuals, particularly in trading and sales roles, can use AI and machine learning tools to make better decisions based on market and reference data.

This is an area with great potential, and Deutsche Bank is actively exploring opportunities in this space in conjunction with its technology and quant teams stated the bank’s Global Head of FIC Quantitative Trading, Roel Oomen.

“Our focus is on AI’s ability to summarise information in a user friendly and bespoke manner,” Oomen explained.

“We are looking to provide the sales and trading teams with tools enabling them to track live market sentiment, generate market colour in a targeted manner, and combine real time news flow with an extensive body of verified research and central bank statements to assist in risk management ahead of scheduled news announcements.”

Oomen stated that the objective is to utilise specialist knowledge of the bank’s staff to filter out inaccurate AI model predictions and, equally importantly, to provide a feedback loop to improve model training. “The human feedback loop is critical for reliable AI development,” he added.

Oomen says the objective is to utilise the specialist knowledge of the bank’s staff to filter out inaccurate AI model predictions and, equally importantly, for them to provide a feedback loop to improve model training. “The human feedback loop is critical for reliable AI development,” he added.

When asked whether there are any obvious downsides to over-reliance on technology in a trading environment, Oomen said that a controlled environment is the main priority. He further stated the objective is to utilise the specialist knowledge of the bank’s staff to filter out inaccurate AI model predictions and for them to provide a feedback loop to improve model training. “The human feedback loop is critical for reliable AI development,” he mentioned.

“But as I have already alluded to, this is about empowering our staff and clients while at the same time use their expert judgment to improve the technology. This re-enforcing loop is key from an efficiency and control perspective, but also from a cultural perspective.”

Many Applications of AI

AI tools have also changed the way financial services companies recruit staff, enabling them to analyse the likely success of a candidate and how they would fit into the culture of the firm by analysing their voice and facial expression to determine their overall emotional state and demeanour.

A report published in June 2023 by Evident Insights suggests there are more than 100,000 employees in the North American and European banking industry engaged in delivering artificial intelligence, with AI-related hires accounting for over 30% of the open job ads at some of the largest European banks, including Barclays and NatWest.

According to Mike Stirton, the Managing Director of the specialist financial service recruitment firm Core-Asset Consulting, there is no evidence of AI being used to replace workers in the financial services industry as yet, but it may not be far away.

“The impact of AI on jobs in finance is embryonic at this stage, but it is coming,” he stated. “While some roles may inevitably disappear, new roles will emerge as a result in areas such as in compliance and cybersecurity.”

This view aligns with the findings of a recent survey by YOHO Workplace Strategy, which found that almost half (47%) of the HR managers and directors of UK financial services firms surveyed needed three years to prepare for the impact of AI on the workplace. Just under one third of the human resource professionals who responded to the survey were concerned about job losses, with 36% expecting AI to lead to a reduction in labour costs.

“AI will displace jobs rapidly in finance and while it will create new jobs, this is unlikely to be at the same rate as the jobs it displaces,” admitted Grant Price, the CEO at YOHO Workplace Strategy.

Artificial general intelligence (which is defined as the representation of generalised human cognitive abilities in software so that, faced with an unfamiliar task, the system could find a solution) could automate order execution and currency monitoring, but it is important to note that some forms of these tasks are already being performed by narrow AI. For example, high-frequency trading firms often use algorithms to execute trades automatically based on predefined criteria, including risk monitoring.

Thomas Friesleben, the Managing Director of the liquidity and trading solutions provider StoneX Pro EMEA reckons artificial general intelligence will become more dominant in traders’ day-to-day conversations. “The challenge will be how to communicate if exceptional market conditions occur, for example when Switzerland announced that it was going to scrap its currency peg of 1.20 to the euro. How would artificial general intelligence react in such a situation?”

He also accepted that while there are many benefits to using technology in trading, there are also potential downsides to over-reliance. “For one, systems can fail – a technical glitch could lead to significant financial loss in a high speed, high stakes environment like FX trading,” said Friesleben. “Furthermore, technology is not infallible and can make mistakes, especially when dealing with complex, unpredictable markets such as emerging market FX.”

The legal ramifications of a machine offering investment advice based on generative artificial intelligence are yet to be tested. While it has the potential to improve access to data, there will be discussions around licencing and sources; for instance, while FX data is readily available at relatively little cost, exchange data is not.

That is the view of David Morrison, a Senior Market Analyst at the broker Trade Nation, who reckons that for now at least most market participants are insisting on striking a balance between utilising technology as a tool to enhance decision making and maintaining human oversight and judgment.

“Over-reliance on technology alone can introduce risks such as algorithmic errors, system malfunctions, and unforeseen market conditions that may not be adequately accounted for by AI models,” he cautioned.

“If any errors get embedded, there is the potential for flash crashes and other dramatic and inexplicable market moves. So, while generative AI shows promise, for now it should be employed judiciously, with appropriate risk management and human involvement to mitigate potential downsides and ensure the overall stability of the trading environment.”

AI tools can narrow the gap between traders with access to Bloomberg terminals and those who don't by compiling and sorting data quickly and easily. However, over-reliance on technology in FX trading may lead to errors in decision making or individuals overlooking better options according to Kate Leaman, the Chief Market Analyst at the FX and CFDs broker AvaTrade.

“It is also important to be aware of the potential for human error, no matter how much technology is being used,” she continued.

Thomas Friesleben, the Managing Director of the liquidity and trading solutions provider, StoneX Pro EMEA believes that artificial general intelligence will become more dominant in traders’ day-to-day conversations.

“The challenge will be how to communicate if exceptional market conditions occur, for example when Switzerland announced that it was going to scrap its currency peg of 1.20 to the Euro. How would artificial general intelligence react in such a situation?”

He also accepted that while there are many benefits to using technology in trading, there are potential downsides to over-reliance. “For one, systems can fail – a technical glitch could lead to significant financial loss in a high speed, high stakes environment like FX trading,” said Friesleben.

“Furthermore, technology is not infallible and can make mistakes, especially when dealing with complex, unpredictable markets such as emerging market FX.”

Exploring the Legality

The legal ramifications of a machine offering investment advice based on generative artificial intelligence are yet to be tested. While it has the potential to improve access to data, there will be discussions around licencing and sources, for example, while FX data is readily available at relatively little cost, exchange data is not.

That is the view of David Morrison, a Senior Market Analyst at the broker Trade Nation, who reckons that for now at least most market participants are insisting on striking a balance between utilising technology as a tool to enhance decision-making and maintaining human oversight and judgment.

“Over-reliance on technology alone can introduce risks such as algorithmic errors, system malfunctions, and unforeseen market conditions that may not be adequately accounted for by AI models,” he cautioned.

“If any errors get embedded, there is the potential for flash crashes and other dramatic and inexplicable market moves. So, while generative AI shows promise, for now it should be employed judiciously, with appropriate risk management and human involvement to mitigate potential downsides and ensure the overall stability of the trading environment.”

AI tools can narrow the gap between traders with access to Bloomberg terminals, and those who don't by compiling and sorting data quickly and easily. However, over-reliance on technology in FX trading may lead to errors in decision-making or individuals overlooking better options according to Kate Leaman, the Chief Market Analyst at the FX and CFDs broker AvaTrade.

“It is also important to be aware of the potential for human error, no matter how much technology is being used,” she said.

Traceability and explainability remain vital in the trading environment. When traders are operating on 'fly-by-wire' over traditional manual order placement, it may be difficult to understand why a specific trade was placed. This is particularly true of convolutional neural networks in which the inner workings of the network are not completely understood.

“In addition, AI software is able to scan large numbers of applicants’ CVs to assess their suitability – a task which would take a recruitment professional several hours to complete,” explained Reece Pawsey, a Director of FinTop Consulting, a specialist provider of recruitment services in the financial trading and trading technology markets.