Get in touch today
To see our real-time advanced analytics in action.
Artificial intelligence (AI) has progressed a long way in a very short space of time, and is set to continue down its evolutionary path just as quickly. FX-MM’s Luke Antoniou strips back some of the conjecture and speculation to find out how this technology is already being used, and how it’s set to develop in the near future…
Artificial intelligence is here, and it’s here to stay, but for now at least, you can close your copy of ‘I, Robot’ without too many concerns about the future’s robot overlords.
Understanding the impact that artificially intelligent technology is having, and will have, on financial services begins with a strong dose of realism; the machines aren’t taking over, but we must face up to the fact that they can perform certain tasks more efficiently than a human being, and just as crucially, at a lesser cost.
AI-enabled software can use flawless logic without a hint of sentiment to make decisions without risk; just look at Google DeepMind’s AlphaGo, which recently defeated five of the world’s best professional Go players, with the world champion explaining that he played the perfect game and still lost. Now, apply that same logic to trading, data analysis and asset management, and you begin to glimpse a future where machines can execute instructions compliantly, efficiently and profitably.
As financial markets become ever more unpredictable, the collection, aggregation and use of accurate data is becoming crucial, and it is that data which is really the crux of this brave new technological world.
The use of predictive analytics, or predictive AI, is helping corporates and banks alike to get ahead in risk management at a time where market movement is increasingly defined by geopolitical risk.
As Mark O’Toole, Vice President of Commodities & Treasury Solutions for OpenLink, explains: “Some great strides forward have been made in predictive AI, to the point we can ask ‘how will it impact me if markets move, or currency moves against me after certain macro events across the world?’ We should now be able to predict a likely outcome, proactively assess that from a risk management perspective, and hedge in advance of that event. Brexit is an example of this, where a lot of companies clearly didn’t anticipate the Leave vote and weren’t doing any hedging of currency.”
While the potential to improve risk management with data analytics is there, it must be remembered that these technologies are still very much in their infancy, as Henri Waelbroeck, Director of Research at Portware explains: “There are opportunities for predictive analytics to assist risk management, for example network analysis, visualisation software or Bayesian network software. However, the field has been slow taking off, not only because there aren’t yet enough compelling solutions, but also because risk groups are focused on old-fashioned methodologies that have become rooted in compliance.”
Naturally, risk management is a vital part of existing within financial services, a fact only emphasised by the global financial crisis and the lessons that banks and corporates were forced to take from it. The technology must be right if it is to take over this aspect of financial security, but once it is, a question must be asked: why trust a human?
“Predictive analytics perform market checks that could be run by humans – the only difference is that machines can run those check rules much faster, and every millisecond,” explains Stephane Leroy, Business co-Founder and Chief Revenue Officer of QuantHouse.
“As time goes on, the power to analyse data signals in real-time and to memorise patterns through AI algorithms makes this whole process even more efficient.”
Automation and the human factor
The relevance of humans in a future where computers will handle monotonous and repetitive tasks has always been on the periphery of the AI conversation, so how can humans adapt in financial services to avoid a severance package?
The key word in all of this is automation, something that humans aren’t suited to and have repeatedly lost out to since the first industrial revolution. The AI argument and use of robotics are intrinsically linked to automation, and there are numerous instances where it should be implemented.
“Robotics and AI mean different things to different people, but the key is automation,” says OpenLink’s O’Toole. “You’ve got a lot of people pulling levers, pushing buttons and spending an inordinate amount of time reconciling databases or spreadsheets when there are already technologies that can take over that work. Automation will allow corporates to better focus on the task at hand, and start to use their heads a bit more when it comes to thinking of the future and what they should be looking at.”
One thing that humans must do to remain relevant is recognise the computer’s function: what it’s doing, why it’s doing it, and the impact it’s having. In doing this, humans move up the value chain to do something that – for now – the machines cannot do themselves.
“Computers are better than humans at processing statistical information, so humans stay relevant by learning how to work with the machines, and how to communicate what the machine is doing to managers who need to know,” explains Portware’s Waelbroeck. “Working with machines is not just using them ‘as is’, but using them to turn ideas into processes: intuition is a valuable trait when it can be turned into a quantitative model.”
There is no getting away from the fact that the fight against automation is one that people will not win, so the real fight is the one to stay relevant in an industry where there will be inevitably fewer jobs.
A vital part of this for people is adapting to new responsibilities and learning new skills in order to keep up and be able to work effectively with their machine counterparts. David Landi, Global Head of Financial Services at The Smart Cube, explains: “Automation through AI tools will definitely have an impact on the number of jobs for research analysts and traders. For instance, an increasing percentage of hedge fund trading is now being performed using automated trading rather than human traders.”
Continuing, Landi highlights some of the skills that people will have to develop to stay relevant once mass automation takes hold of the industry, and how job roles will change rather than disappear: “AI isn’t going to wipe out the jobs of analysts or traders, but their roles may change in the process, and therefore, the skills they would need to possess to grow in this market space will need to evolve. For instance, quant skills – using computer-driven models for trading – data science and computer science may become an important knowledge to hold in the industry.”
The use of AI and machine learning is also proving helpful to compliance teams, not just freeing up time and resources for an organisation’s staff through the automation of monotonous tasks, but ensuring that firms are being compliant even quicker.
Financial software provider Misys has recently made the foray into compliance-based AI solutions with a trade monitoring platform, for example. Its new solution red flags probable mistakes that can prove crucial to EMIR and FRTB compliance, as EMIR requires trades to be confirmed without errors within 48 hours, and FRTB requires daily risk reports which would breach compliance in the event that unidentified trade errors are found.
Indeed, as regulation continues to tighten up across Europe, the chance for firms to make compliance as simple and as inexpensive as possible is one that they will jump at.
“Compliance processes are vitally important and need to be done with high precision, but they involve a lot of standardised procedures. These are exactly the sort of tasks which are ripe for automation,” says Matthew Hodgson, CEO at Mosaic Smart Data. “AI allows these repetitive tasks to be automated, leaving staff to focus on the more subtle, valuable and interesting work.”
As there always is with the adoption of new technology, there is a timescale and schedule to consider. While the future may not be here just yet, the pace of technological evolution is so rapid that it’s unlikely to be long before AI and automation dominate standardised tasks.
“I don’t think AI will be an immediate threat to the jobs of traders, but it will be a threat to the basic and repetitive tasks carried out by traders,” says QuantHouse’s Leroy. “Of course, if some traders are focused on such basic and repetitive tasks, that could become a threat to their role within the organisation.”
A part of what makes AI such a tempting proposition for financial organisations is the fact that machines do not cause problems or take risks that could cost the company money in the same way that a human might – as Leroy notes, “technology is a solution, not a risk”. With that said, though, there is a risk attached to it, and it does revolve around cost.
A question of cost
While AI doesn’t cost FIs money in the way that human error could, implementing the technology represents a fix that won’t necessarily come cheap. This is something that banks and corporates are learning again, as they always must when the necessity of a new technology rears its head.
Indeed, new technology almost always comes with something of a hefty price tag, be it a new smartphone or a piece of AI software. Cost is something that financial institutions must always be wary of, especially when they’re weighing up whether or not to build their own technology or adopt somebody else’s.
“Cost has become a crucial factor in any decision related to technology,” explains Mosaic’s Hodgson. “Banks are constantly looking for ways to eliminate risk from business operations, but the unpredictable cost and resource demands of in-house technology projects have contributed to an increase in the levels of outsourcing.”
Keeping hold of top talent could also be an issue when it comes to building in-house AI systems, says Hodgson, and it’s a theme that banks will hope is not recurring; a survey from Synechron earlier in 2017 showed that a talent gap in blockchain development could be preventing the technology from going mainstream. In the same way, banks building AI in-house will be hoping that they don’t lose their best IT developers, thus creating their own talent gap.
“Another frequent problem with building infrastructure in-house is that, if the IT staff in charge of the project are brilliant enough to make it succeed, they tend to move on to other things in a couple of years,” continues Hodgson.
So, what is a bank to do? The answer, as it always seems to be at the moment, is to collaborate.
“Partnering with an external provider will give a bank much better continuity of support and development of the product, which is why partnering with fintech providers is such an effective model,” says Hodgson. “Fintechs have already incurred the cost and resources of building technology, and are best able to efficiently support and enhance it. By opting to outsource trading technology and paying attention to their core business, institutions are able to quickly deploy cutting edge systems.”
Indeed, bank and fintech collaborations are springing up in conjunction with almost every area of financial services, with the agility of the latter often sparing the former a lot of time and money. Naturally, not everybody agrees, with The Smart Cube’s David Landi explaining that AI and machine learning will “remain a collaborative effort for small- and mid-sized players”, but that the bigger players seem more inclined to develop new capabilities in-house.
Whether firms collaborate or go it alone, AI is beginning to make waves in financial services, and increasingly looks to be the beating heart of a bright future where risk ranks far lower down the list of concerns for FIs.
At the centre of it all, though, is the human race: whether it is the developers that create new software, and give chatbots the peppering of personality that make them relatable, or the executives that provide the spark to drive an organisation’s future, ultimately, the question isn’t one of human relevance, it’s one of human imitation.
Junior Editor, FX-MM