The buy-side on AI: ‘The fear is real, but the rewards are there’

AI is really changing just how much a trader can add to the investment process. Not just through execution, but also the selection aspect of what goes into a portfolio,” asserted Rebecca Healey, co-chair EMEA regional committee at FIX Trading Community.

Fifty industry experts joined the recent FIX Trading Community London Artificial Intelligence (AI) Workshop voicing – among other views – that the upsides of AI are there despite some existing fears.

Speaking in a roundup of the findings, executive director of FIX Trading Community, Jim Kaye, explained that “the room didn’t feel that the industry was quite ready yet for full AI in autonomous pre-trade and execution solutions, but did say that it was only a matter of time.”

The insights, gleaned from a set of industry experts across three roundtable sessions, broadly address the use cases for AI in trading and implications for data as well as touching on risk and mitigation considerations.

The discussion groups were largely made up of traders and data experts, specifically: 17 buy-siders, eight sell-siders, 19 vendors and eight other parties including regulators.

Kaye highlighted that due to the broad mixture of people in the room, there were a fairly broad set of answers to each topic.

AI, a veneer for progress

One key finding from the room was the assertion that for AI, an effective way to encompass its potential would be to view the technology as an overlay to existing processes, rather than as a replacement.

“An interesting point made about legacy systems was that the industry is not necessarily trying to fix them as such or replace them, but simply understand them better,” said Kaye.

“The idea of having some autodocumentation of what legacy systems do and how they work allows for better support, it’s about putting layers of technology on top of these systems to make them easier to deal with rather than having to potentially throw them out.

Read more: Fireside Friday with… FIX Trading Community’s Jim Kaye

Overall, the strong sentiment from the workshop was that for the meantime looking at the short-term future, it remains all about improving what humans can do rather than replacing them. Helping make human touch more proficient. 

Speaking to this, Rebecca Healey, co-chair EMEA regional committee at  FIX Trading Community asserted that groups agreed that AI is truly adding to traders’ effectiveness, working to enhance performance: “It’s really changing just how much a trader can add to the investment process. Not just through execution, but also the selection aspect of what goes into a portfolio.”

She further added that, when it comes to TCA, it is becoming clear that what has been able to be done in equity, the market can now begin to do in other asset classes.

“It really creates some interesting opportunities, particularly around liquidity profiling and instrument selection.”

Data then AI, AI then data?

When it came to what the workshop thought about the relationship between AI and data a large part of the discussion centred on how structured data is or should be. 

Matthew Coupe, co-chair EMEA regional committee at the FIX Trading Community, shared that when it came to unstructured data and structured data there was an element of disconnect within the discussion group: “A disagreement if you like. There’s a general sense that you need both of these and can do different things with the different types.”

Coupe highlighted that while the importance of structured data in order for artificial intelligence to be effective in trading applications is true, AI is getting better and better at handling unstructured data.

“There’s a general sense in the room that you do need the data completely structured and to some extent, normalised to be able to use AI really and let it loose on trading type applications.

“You need to find what standards are obviously and then you need to do some work to clean the data up, and this is an expensive and time-consuming thing. But if you don’t do it, then you basically can’t trust the results, that was generally the sense.”

Interestingly, one discussion point raised was the notion of AI as a cleanser and producer of data, as opposed to simply a user. Something which could be done up front, and potentially take priority.

“This is an interesting area for the industry to really look at, how can we use AI to help improve what we do to then use the AI tools in a better way going forward,” asserted Kaye.