At its most basic, artificial intelligence is all about getting computers to do tasks that would historically require human intelligence. In turn, machine learning is about building the machines that apply algorithms to reams of data, which then learn to identify patterns and insight in the data. That ability to learn, and to do this at scale, has had big impacts in many sectors, not least in financial services.
The changing face of markets
The financial industry has had a tumultuous decade. Trust in financial institutions remains fragile, while intervention by regulators has created new burdens for firms to contend with in a challenging economic environment.
Venkataraman Balasubramanian (Bala), chief technology officer of banking and capital markets at IT services group DXC Technology, says the focus on artificial intelligence and machine learning differs somewhat across regions. Where cost reduction has been a priority in Europe and North America, technology has been used to automate iterative processes.
Yet advancing technology also has the benefit of helping to level the playing field in the financial sector. As the World Economic Forum outlined in its report on ‘AI and the physics of the financial sector’, financial organisations in the past could prosper - partly through size and huge economies of scale, offering standardised products.
This gave them unrivalled access to markets and helped them to build exclusive, longstanding client relationships.
But the competitive landscape looks very different today. Scale still matters, but it is now important to have in terms of data, and the technological capacity to deal with it.
There are opportunities for smaller firms not just to disrupt the market but also to provide their entrepreneurial mind-set to more established corporations. Pedro Porfirio, who runs global capital markets at Finastra, argues that emerging markets such as in the Middle East are more prone to disruption and that there’s a wealth of opportunities for smaller businesses with the most interesting innovations.
Using technology to catch criminals
As well as democratising access to financial markets, artificial intelligence can go some way to repairing some of the trust issues in the sector. Machine learning has the advantage of being able to look through huge volumes of data and spot something that doesn’t fit usual patterns. That provides banks with a crucial ability to better spot criminal activities such as fraud, money laundering and market manipulation. Whether analysing factors such as transaction volumes, the jurisdiction or destination bank account, machine learning allows computers to flag suspicious transactions. This goes further than simple rules-based systems, which can generate multiple alerts that are time-consuming to investigate. Using machine learning together with the vast data sets available, banks can train the computers to prioritise suspicious activity, allowing the compliance staff to better focus their time and efforts. All this reduces dependency on people to perform time- and resource-intensive tasks.
“From the standpoint of audit and compliance, machine learning is an incredibly powerful tool”, says Porfiro. “The ability to spot market manipulation, whether that’s front-running or insider trading, can be greatly enhanced through machine learning.”
The exchanges, such as Nasdaq, have been using machine learning surveillance to detect unusual trading activity for several years.
From risk to revenues
While the ability to spot fraud is one of the huge advantages of artificial intelligence, the technology also has the potential to improve the bottom line. Banks have created platforms that can analyse legal documents and extract key clauses or data in seconds – a task that, done manually, would take hundreds of thousands of hours. Such tools mean employees’ time can be spent on more value-added work for clients.
Bala thinks the revenue-generating potential of artificial intelligence is significant.
“For me, the biggest upside in artificial intelligence will come from providing clients with more bespoke trading ideas,” he said.
That could mean using the technology to help identify value in stocks or using it to profile customer data alongside previous decisions they’ve made in order to tailor investment advice.
“After you buy something on Amazon, the algorithm shows you what others have bought,” Bala explains. “Artificial intelligence can do a similar thing with stock picks.”
By better understanding trading preference and behaviours of clients, companies can make a more informed decision about pricing and risk appetite for the client.
One of the fears about the prevalence of artificial intelligence centres on humans being replaced by technology in the workplace. Indeed, one of the most prevalent uses of artificial intelligence and machine learning in financial markets is chatbots. While the most basic level of customer interactions, such as loan queries or account checking, can be completely automated, the more complex and value-added queries require the chatbot and human to work together. The chatbot can easily ingest an enormous amount of data around - for example, a bank’s offerings or market insights - and can assist the call centre staff allowing an accurate, and consistent, customer response.
Porfirio agrees that artificial intelligence can augment what people do, rather than replace it. He points out that in market-making for example, where a trader makes a decision to buy or sell at certain prices based on various information, a machine-learning algorithm can use the past behaviour of that trader to make its own buy and sell decisions. This would allow traders to scale up the volume of transactions and better transfer risk.
“The interesting question”, notes Porfiro, “is when the machine can detect patterns better than humans and when artificial intelligence can replace human intuition.”
Algos and the flash crash
From spotting fraud to picking stocks there seems to be plenty that artificial intelligence can offer. But what about the risks? The extreme volatility seen in financial markets has been attributed to the increasing use of automated trading strategies. The theory is that these create a herd mentality, exacerbated in times of low liquidity. As equities or currencies move lower, it can trigger more sell signals. Some algorithms are also designed to feed off news headlines or social media, meaning the potential for an additional downward spiral.
Bala is not convinced. “I haven’t seen any evidence that algorithms are causing the crashes,” he argued.
Porfiro agrees, explaining that the biggest issue with some of the ‘flash crashes’ that have taken place – where an asset massively decreases in value, for a few seconds or minutes, is people using dynamic hedging to hedge a very large number of equity options.
“Using more artificial intelligence and machine learning will raise a flag if it hasn’t seen a pattern before. It’s different from hard coding so you could argue that it actually makes the market more resilient. You will always have herd behaviours in markets, particularly in times of stress but these flash crashes existed long before the algorithms came along.”
Porfiro argues that a couple of the big transformative effects that could take place in capital markets will be around the use of blockchain to engender trust between capital market participants. It could also have a major impact on financial inclusion, he said.
“It’s been said we overestimate the impact of technology in the short-term and underestimate it in the long-term”, Porfiro says. “Artificial intelligence is here to stay and that’s a good thing.”
(Reporting by Charlotte Kan; Editing by Michael Fahy)
Our Standards: The Thomson Reuters Trust Principles
Disclaimer: This article is provided for informational purposes only. The content does not provide tax, legal or investment advice or opinion regarding the suitability, value or profitability of any particular security, portfolio or investment strategy. Read our full disclaimer policy here.
© ZAWYA 2019