Exploring how algos are becoming increasingly central to FX trade workflows

Algo sophistication is a major reason for this shift. Gordon Noonan, Head of FX Trading at Schroders notes that the algo space has evolved considerably over the past decade. “Algorithms have moved from rudimentary TWAP orders to more intuitive strategies designed to adapt to prevailing market conditions,” he says. “We have worked closely with bank algo developers over the years to understand how algos operate in the FX market, the logic each algo utilizes, and what venues they access. Our equities team have been big users of algos for a number of years. The FX desk were able to use their extensive experience and research methodology to monitor our algo execution internally.”

“The growth of third-party independent TCA providers allows us to assess individual algo execution across several metrics,” he notes. “This also allows us to compare different algos that follow the same execution strategy.”

Macro events helping algos replace traditional workflows

“Comparatively speaking, adoption of algos, within the buy-side for the FX market, is still significantly behind other major markets,” says Colby Jenkins, Strategic Advisor at Aite-Novarica Group. One of the reasons for relatively slower uptake was the inertia caused by existing voice trading workflows. Simply put, traders found it easier to call stakeholders and get a feel for market depth and color.

Meanwhile Elke Wenzler, Head of Trading at MEAG, notes that voice will retain some utility in the FX world, although technological innovation is quickly changing that picture. “Previous crises have shown that despite all the achievements we have in electronification, having an adequately staffed trading desk is essential,” she says. “This helps us switch very quickly to high-touch in very volatile or even broken markets, and have sound knowledge of market structures and their main market participants.”

Wenzler cites the example of the current conflict in Ukraine as a situation where switching to high-touch was essential. However, she also says that algos are beginning to fill the space that voice used to dominate. “With the advancement of e-trading and algo desks, many banks have specialists in the electronic market that we have interacted with before, during, and after algo execution,” she explains. “This has provided us with some of the information that we used to get from the voice desk.”

Wenzler notes that algos offer more control over order execution style. Schroders’ Noonan echoes this sentiment. “Algos provide buy-side traders with another instrument to provide the best execution for our clients,” he says. “They allow traders to take direct ownership of order execution, giving us increased visibility of prevailing market conditions. Traders can use this information to tailor our execution to improve overall execution costs.”

Aite-Novarica Group’s Jenkins offers a broader context when asked about the factors behind the rise of algos in FX. “There is no question that the market turmoil we saw in early 2020, coupled with work from home mandates significantly pushed the electronification of many markets to new heights,” he says. “This was particularly pronounced in certain corners of the fixed income markets, and certainly was the case with buy-side FX traders.”

He notes that any concerns about these trends being temporary have vanished. “It seems clear now that there is a growing appetite and a widespread market expectation of further growth,” he notes. While global events have created space for algo usage on the buy-side, trader adoption is increasing because of the utility algos offer.

So how is the buy-side using algos, and what are some of the advantages on offer?

Spot execution, risk management, and efficiency gains highlight algo advantages. Execution is the first algo use case Noonan offers when quizzed about utility. “We look to use FX algos that balance the tradeoff between time risk incurred during execution versus savings from spread capture throughout the execution time horizon,” he says. “We generally find that POV/float algos provide that balance and suit our trading style best.”
Unsurprisingly, workflow efficiency is one of the primary benefits that algos unlock. Automation throughout the pre, in, and post-trade windows has resulted in better pricing and lowered risk exposure. While algos aren’t a cure-all for the risks firms face, they are undoubtedly serving a purpose.

Wenzler offers a specific risk management example. “A prime use case for us is strategic adjustments of large hedges such as those driven by ALM management,” she says. “We use algos to minimize the risk of large market impact during execution. We select an appropriate algo depending on the type of risk we wish to manage.”

An algorithm’s ability to quickly process large, complex datasets and offer insights is unparalleled. Jenkins notes that while these characteristics offer advantages, a lot comes down to the data firms feed into algos.

“As the technology develops, and the algos can reliably take in increasingly complex sets of data- be it size, liquidity characteristics of the pairs, or when the order was placed, and so on- the application of the tools should naturally evolve from the simplest use case to more customizable applications,” she says.

Wenzler agrees with the sentiment that algos are rapidly increasing efficiency in trade workflows. “Algos have allowed us to react faster, replicate, and extend trading strategies,” she says. “We can use our resources more efficiently to concentrate on other tasks to add more value to the desk.” She also states that automation has reduced the need to be hands-on with small trade tickets.

Algo usage in the FX spot business has grown considerably and continues to witness the most development. The highly-custom nature of order flow in the NDF business makes algo adoption challenging. However as technology improves, there’s no doubt that algo use cases will rise in that arena.

“We see scope for the growth in NDF and swaps algo trading in the future,” says Noonan. However, he notes that there is still considerable development to be deployed in the spot space. “We believe that despite algo execution in the deliverable spot space maturing throughout the last few years, there is still scope to provide real-time, inflight execution feedback, detailing how the algo is interacting with the market at that moment in time,” he explains.

Wenzler echoes these views. “G10 Spot – main crosses/spot – is our focus right now,” she says. “Forwards are another topic, but we also see the advantages of (algo use in) NDFs.” Wenzler notes that the possibility of tapping internal liquidity or flows in NDFs is lesser when compared to spot. As a result, algo development in this area still has some way to go, but has already improved significantly over the past year.

Further discussion of algo usage in NDFs and swaps reveals that there is still some work to be done by service providers. While the buy-side is enthusiastic about algo adoption, algo service providers must take a long list of desired features into account before deploying future solutions.

Where algos can improve

Despite her enthusiasm for algos, Wenzler notes a few areas where they currently fall short. “One area of improvement we would like to see is the building of dynamic rules that take into account current market conditions per currency pair, such as liquidity and volatility and supports the traders in choosing the best set of trading strategies and algo combination pre-/ in- trade,” she says. The complexity of FX data is the major hurdle in this case. Wenzler hints at leveraging machine learning to help with this cause.

“We are in the very early stages of looking at ML and the idea is to see if it can help determine/advise on the best way to execute orders in different situations (e.g how many banks to ask, how automated should the execution be, etc)” she says. She notes that while some algos use ML currently, MEAG is focusing on leveraging them during pre-trade processes.

Noonan also believes in AI and ML’s potential for algo execution. “The continuing improvements in AI and machine learning have the potential to improve algo results in the future,” he notes. These technological advances will help algos adapt to market structure faster. In turn, faster adoption will offer buy-side traders more color on market depth, perhaps even replacing the need for voice trades.

However, AI and ML are in their early days, but rapid progress is underway. Aite-Novarica Group’s Jenkins is seeing buy-side demand in FX mirroring those in fixed income markets. “It looks like the buy-side more than ever wants more control in their hands concerning how the algos are deployed,” he says. “The analytics driving them are also important,” he explains that buy-side traders want greater visibility into what is happening beneath the hood.

Pre-trade workflows are a fertile area for further improvement and tie in with Jenkins’ point about enhanced transparency. As Wenzler explains, “Further integration of pre-trade analytical tools from third party providers or banks used within multi-dealer trading platforms does enhance the experience further for market participants rather than having to sign into different single dealer algo suites to accomplish it.”

She notes that MEAG’s business needs and the current lack of integration pose challenges to workflows. “For some orders, we need to get in or out of risk very quickly, and would be using implementation shortfall strategies to beat risk transfer prices,” she says. “But this usage also needs to take into account the impact on the market for the size.” Jumping between multiple platforms to assess pre-trade risk is an obvious flaw in this scenario that has to be solved.

TCA

TCA is another area where algos offer benefits but have immense scope to improve. Given the way the markets are evolving, traders of the future will likely integrate automated TCA into transactions that do not need special attention and focus on the ones where they can add value.

Needless to say, this talk of automation and integration makes transparency even more important. Visibility into code and logic will help firms choose the best algorithms to deploy, given certain market conditions, pair dynamics, and firm needs. This visibility will also help firms adopt a hands-off approach to their trading more often, something that is critical to greater algo adoption. Jenkins believes that TCA standardization will boost algo adoption on the buy-side.

“Putting a finer point on it, more standardization around TCA is certainly on the horizon for FX,” he says. “We can expect to see the greatest uptick in algo adoption in segments of the market that currently underutilize algo tools such as NDFs and the less liquid markets.”

What lies ahead for algo usage in FX?

Noonan, Wenzler, and Jenkins are highly optimistic about the continued growth of algo usage in FX. Wenzler notes that algos proved their worth by helping firms manage execution during high volatility periods in the market.

Jenkins says, “I would expect the market to continue along the current trajectory of algo adoption within the buy-side. Across the board, we are seeing improvements in granular trade data as markets shift to electronic structures.” This shift creates a virtuous cycle where data usage increases sophistication, which in turn necessitates the use of advanced technology such as algos.

While no one can accurately predict what the future holds, there’s no doubt that as algos become more sophisticated than ever, their usage in non-traditional FX business areas will increase. In turn, the buy-side will undoubtedly have new needs and risks to mitigate, prompting even more sophistication within algos.

In short, the future for algos in FX is promising and bright.