Multinational tech giant IBM announced in June that its Watson supercomputer had been enlisted to help run an exchange traded fund (ETF) by digitally  selecting the stocks that would out-perform the US market index. It was an eye-catching technological development but – in the bigger picture – nothing new. Automated computers and artificial intelligence (AI) have been used in high-frequency trading since at least the late 1990s, when they astounded traders around the world by executing trades 1 000 times faster than humans.

‘It has grown in leaps and bounds since then,’ says Xolani Holomisa, the JSE’s Business Intelligence Manager. ‘With the explosion of computing power and big data technologies, it is now possible to collect real-time or near real-time structured and unstructured – social media, images, video, voice – data, and have it ingested by machine learning algorithms.’ Those algorithms are able to crunch millions of data points in real-time – a process that was not possible in the old statistics models, which only worked on sample data.

Now, as machine learning, big data and highly advanced internet of things capabilities become more of an everyday reality, AI is starting to show real value as a stock exchange tool.

The JSE is currently exploring a range of applications all along its value chain. ‘We are in an exploratory phase at the moment,’ says Holomisa. ‘We’re looking at putting together a data science competence for our internal customers, and externally we are looking at various delivery mechanisms, including partnering with some of our customers as well as consulting firms that are already ahead in this space. We are also actively looking at various big data analytics and cloud solutions.’

The opportunities, he says, are endless – both as a value-add to JSE members and listed companies, as well as in terms of possible revenue streams. AI also has tremendous value for business intelligence (BI) – data analysis that provides past, present and predictive views of business operations. ‘The current BI technologies help us understand what, when and why things happened,’ says Holomisa.

‘AI will put us in a better position to accurately predict what is likely to happen next, and also prescribe to us how we should handle certain events when they do occur. We will be able to build intelligence alerts that will inform us as soon as something unexpected happens or when a certain threshold is met using a wide variety of data from disparate sources.’

Traditional BI relies on gathering insights based on historic data, but the pace at which the markets operate makes it impossible to make key decisions based on this information. AI will enable the JSE to conduct real-time data analytics with actionable insights, all in real-time. ‘Due to time and resource limitations, and the large volumes of the data we analyse daily, we are currently unable to get our hands on all available data that the JSE and other markets generate, so we might be missing out on key insights,’ says Holomisa. ‘AI will enable us to quickly sift through all that data and prescribe key data elements that can prove invaluable in understanding or solving complex problems.’

While this level of data processing is achievable only with AI, Holomisa doesn’t see humans being removed from the BI process completely. Not yet, at least. ‘There will always be a need for a human,’ he says. ‘Especially in the field of data, where there is often a lot of ambiguity, data quality issues and perhaps even data interpretation issues. The role and skills set of the future BI analyst will certainly change as AI technologies mature, but I don’t see AI completely replacing the human mind.’

By Mark van Dijk
Image: Gallo/GettyImages