Embracing artificial intelligence for the buy-side

The increasing use of artificial intelligence and machine learning systems among buy-side firms is in danger of creating a hype bubble – The TRADE examines where asset managers are currently using these technologies to optimise their trading strategies and what pitfalls firms must avoid to (eventually) foster a harmonious trading between man and machine.

The application of artificial intelligence (AI) and machine learning (ML) technologies in financial services is being increasingly positioned at the vanguard of technology-focused industry discussion and strategies, as discussions focus on the practical implications of deploying such tools beyond middle- and back-office functions.

As margins are squeezed and costs continue to rise in light of fee compression and an increased regulatory environment, the buy-side is increasingly turning to AI and ML in the search for further efficiencies.

There can be no doubt that huge increases in data and trading volumes that the buy-side is now transacting, is forcing the need for new technologies. While taking part in a panel discussion at this year’s TradeTech Europe conference, AXA Investment Manager’s global head of trading, Daniel Leon, told delegates that his firm has no choice but to invest in new technologies because his trading desk simply cannot keep up with the sheer amount of data and information required to maintain its trading activity.

“We are not able to do what we used to do 20 years ago,” said Leon.  “Yes, you can have a specialist on leverage loan, but on the big credit market or medium and small-cap you cannot have all that information on one guy. We are trying as well to solve problems that we used to do a long time ago. For the more vintage traders it used to be that the trader would know the market and what’s traded for one month, what happened last week, they had information and that’s what typical trading used to be.

“But now we have to gain efficiency, we have to trade so many bonds that you can’t ask one trader to remember everything, to know that this sector last week had this event. We have to reconstitute the experience that the trader used to have: What has traded, what was the liquidity and what was the market impact. You can’t do that on a comprehensive basis.”

BlackRock’s global head of trading, Supurna VedBrat, echoed Leon’s sentiment on the importance of AI and data for the future of the industry at this year’s US Fixed Income Leaders Summit in Philadelphia. Focusing on fixed income markets, VedBrat told delegates that not only will AI be a key element in the next evolution of buy-side trading operations, but it will likely morph the role of the buy-side trader in the process.

“Data science and AI give us the ability to truly augment human intelligence with computing power, and you are able to do that at scale. I think it is going to materially change trading strategies that the buy-side uses. You don’t need human intelligence to pick trades, so you can automate a lot of that flow and the trader is now much more of a risk manager overseeing that the market is working the way we expect, and if not, they have the ability to step in and correct it,” VedBrat said.

Ahead of the curve

Research from TABB Group earlier this year has, in fact, suggested that the buy-side is slightly ahead of the curve in terms of AI adoption compared to the sell-side and exchange operators. Over 80% of asset management respondents stated that they were at least in the planning or research phase of implementing AI, compared to 73% of their sell-side counterparts and exchange operators. At the same time, more than 60% of buy-siders said they expect spending on AI to increase over the course of this year.

According to the research, the majority of asset managers agree that actionable insight is the biggest benefit of deploying AI technology, followed by increased efficiency and automation, strategy selection and risk management.

However, there is a false perception can sometimes be that AI and ML are relatively new to institutional trading; the truth is that both buy- and sell-side organisations have been exploring, developing and implementing such technologies for many years now.

“The key takeaway from all of this is that most capital market participants are bullish on the use of AI and big data in the near future. It is high on the change agenda at most firms, with the main use case being around the investment process, but also in trade execution and operations,” the research from TABB Group concluded.

As with most technology trends though, hyperbole has a way of dominating the discussion. Similarly to the way blockchain exploded into the financial markets’ consciousness in 2016, AI and ML have become industry buzzwords, or at the very least a misleading shorthand, that risks overstating practical applications.

Ian McWilliams, investment analyst at Aberdeen Asset Management, detailed how the understanding of what ML technologies are capable of is being distorted by a lack of understanding and exaggeration, during a panel discussion at TradeTech FX Europe at the end of last year.

“I joke that when you are advertising externally you say AI, but inside you say machine learning and actually you are just doing logistic regression and things like that,” he said. “I don’t think that’s disingenuous, maybe it’s a bit of hyperbole, but it’s not wrong in terms of definitions, because when we talk about machine learning it really is anything where you are getting an algorithm to learn from data.

“We’re taking a lot of market signals and sentiment signals, forecasting what markets are going to do in the future and using those to build trading strategies.”

McWilliams explained that the hype around elements of ML such as deep learning, image recognition and natural language processing (NLP) are distorting expectations around what are essentially tools to better model data for trading strategy decisions, particularly when it comes to conversations with fund managers.

“The interesting thing we need to think about as an industry and maybe where attitudes need to change is around interpretability of the models, which is a big question in a lot of areas, not just finance,” he said.

“Whenever we come out with a trade a question we get asked by the traditional fund managers is ‘Why is it making that trade?’ and they generally expect a very causal, A to B explanation, but that often defeats the point of these very complex algorithms. The middle ground is not good enough to just say that the algorithm says to do it, so we are doing it, but there needs to be more conversation between the quant people and more traditional people to understand there is a trade-off there.”

Beyond the middle-office

As asset managers continue to experiment with AI and ML, the goal has always been to automate manual and often repetitive tasks for greater efficiency and cost savings, freeing up time for traders to focus on more pressing tasks or complex order flow.

But, according to market participants and technologists, the use of AI and ML elements are now permeating into more intricate parts of the business. AI and ML are beginning to show value when it comes to pricing and seeking liquidity, challenges that are often highlighted by buy-side traders in the current market conditions.

“The simple trade automation, the idea of creating rules to take some of the more liquid or easier to trade orders off the books, makes sense,” said Ian Mawdsley, head of buy-side trading for EMEA and APAC at Refinitiv, during a webinar hosted by The TRADE in March.

“The reality is that we have been using both of these processes [AI and ML] for some time. If we look at algo trading supplied for the sell-side in particular, much of that was formed in the first place to automate some of the more menial tasks sales traders were performing. That has now been taken to the next level where people are looking at price discovery and liquidity discovery.”

Further to this, looking at the practical applications of AI and ML, an area that has been of particular interest to the buy-side is the algo wheel, or broker selection processes. While an algo wheel is technically a form of AI, it is on the more basic, rules-based end of the spectrum, but it does provide a solid foundation to build upon.

JP Morgan Asset Management has homed in on this space and produced a framework, known as STARS (Systematic Trading Algorithm Recommendation System), which aims to optimise the way in which traders choose algorithms using ML technology. According to the firm’s global head of equity trading automation and execution, Ashwin Venkatraman, the vast amounts of data now accessible in the market underpins and is at the heart of implementing these new tools on the trading desk.

“We’ve had [STARS] since 2017, we’ve had 90% of our algorithm placements, even back then, going through the framework and we are rolling it out globally as well,” Venkatraman said on the webinar alongside Refinitiv. “In many ways we have been there and we’ve been optimising that wheel over time. We are trying to think about this more holistically. It’s really about being data-driven in the sense that wherever we look at all functions of trading, there are different aspects to it and it is about trying  to leverage that data in the most appropriate way.”

Taking part in a keynote discussion at this year’s TradeTech conference, Antish Manna, head of execution research at MAN GLG, said that the firm went live with a machine learning-based framework for order flow and broker allocation last year.

“This framework effectively takes away the need for human to set an arbitrary target for ‘my first three brokers are going to get this amount of flow’ and continuously updating that target to having a machine that automatically does that”, Manna explained.

“The beauty of it is that it becomes a very clean conversation with our brokers; they know how we are doing things and that they will get more flow, and this machinery also adapts to changing market conditions.”

Man and machine

Despite all of the potential benefits that may be realised, there are significant obstacles when it comes to deploying AI or ML processes, mainly in the form of compliance hurdles, transparency concerns and the build vs. buy dilemma that most firms will consider at some point during implementation.

Firms are urged that they must engage with compliance departments when undertaking any technology project, and the importance of continuously assessing the model to overcome some of those transparency barriers is paramount.

The adoption and successful use of AI or ML comes with a significant resource cost attached, and as such, firms that expecting to realise quick results will be sorely disappointed unless they are prepared to play the long game.

Addressing these unrealistic expectations, MAN GLG’s Manna said that the majority of time spent on machine learning projects is used to clean data before research and development can take place, and that those firms that are only now starting their journey with machine learning should be not expect to see results in the short-term.

“The truth is, it is a fallacy and it takes a huge amount of time to build a framework where you can deliver things at scale that work,” he said. “On the machine learning and AI side of things, problems are best solved by teams of people, because you need the challenge, rigour and time to learn and fail, learn and try again; that process takes a lot of time.”

Another significant challenge is in finding the right balance between man and machine, as market participants and technologists attempt to dispel the myth that AI and ML is even close to replacing the human buy-side trader.

“Technically there is a bit of truth that everything could be automated, there is no doubt that most processes could be run without a human being, certainly within our industry,” Mawdsley explained. “The point is that the world isn’t that flat and there are certainly unusual things that happen in life every day that don’t follow the patterns and I am not sure that we are 100% there, where the AI is able to interpret all of those black swan events and build them into a model.

“There is an element that says the human brain is trained to deal with these outliers, and the machine giving help to follow those patterns is probably where you want to be… It’s about using technology to make more informed trading decisions and as such that means not fully automating everything at all and ensuring that we are bringing in an element of human intelligence, but we are presenting people with options.”

As data becomes more readily available and the size of that data continues to grow, there is little doubt that asset managers implementing AI and ML technologies are at the forefront of the future of buy-side trading, and this is happening now. As some funds struggle to adapt to the changing trading landscape, others are ready and willing to seize the opportunity using AI and ML to overhaul traditional trading processes.