What’s influencing FX execution algo product development?

The issue of FX algos being able to source high-quality liquidity in a fragmented liquidity landscape is a long-running topic and continues to be a challenge for many buyside participants, says Alexis Laming, FX algo trader at Crédit Agricole CIB (CACIB). He explains at CACIB, however, all algos are designed to be able to cope with these liquidity issues and to be able to navigate this landscape. “Our clients are using algos and are seeing algos as a tool helping them cope with liquidity fragmentation,” he adds. “We offer customised liquidity pools and can design ad-hoc customisations of those liquidity pools which overcomes many liquidity issues, including concerns about liquidity mirages.” At the same time, Laming warns that there is not the case that ‘one-size fits all’ when looking at algo strategies. When clients face the trade-off between minimising market impact, reducing market risk and maximizing execution certainty – the overarching algo execution ‘Trilemma’ – algos are useful but should still be recognised as a tool in the trading toolbox, rather than a ‘magic wand’. 

The use of algos alone cannot solve this Trilemma, so the focus needs to be on the client and their expected execution outcomes, Laming explains. “Our focus is always on engaging with clients and understanding their needs,” he adds. “And eventually, we can look at customising the tools by amending the trading logic, or we can design very specific liquidity pools for an individual client, or even add or remove some of the parameters that are embedded in the algo. It is similar to Formula One, where even with the same engine, individual cars may have very different outcomes. A small design change for the car can change a lot in terms of performance. The same principal applies with algos. Everything comes from having these discussions with the client first, engaging with the client, understanding their needs and then working towards meeting those needs as well as possible.”

As a result, collaborating with clients is an essential part of how CACIB operates, Laming continues. “They are the users of our tools and we are completely aligned with our clients to ensure they have the best execution outcomes possible,” he says. According to Laming, this is why regular engagement with clients and listening to client feedback or providing analytics data to help work towards achieving their expected execution outcomes is key. “At CACIB, we have a very wide client base, with some corporate clients who use algos only around once a year, through to our very sophisticated clients who are now executing with algos on a daily basis,” he adds. “It is not about the size of the order being executed, but how well we are able to work with them to understanding their trading needs and adapting our tools to help them achieve these goals. Our algos are set up in a way that allows us to tailor the algo parameters to suite the client.”

Changing execution priorities

The necessity for data insights and feedback has also been elevated due to the uncertain market conditions, prompting conversations about whether the parameters and strategies that worked during more predictable markets are suitable for these volatile periods, says Preston Mesick, Global Head of FX Algos at Barclays. “From the algo perspective, this has meant looking at all the data with customers and with the TCA providers to make sure we are investing in enhancing the areas which will prove to be the most valuable for our customers,” he adds. 

Ajay Kataria, Head of Electronic FX Distribution, Americas at Barclays, agrees, adding that on the sales side they have also been actively helping clients to navigate different volatility and liquidity regimes, especially when using the algo suite. “This is not a ‘set it and forget it’ market. We’re dealing with headline risk on fly,” he adds. “Being able to help our clients utilise our tools effectively, utilising their limit prices effectively and then choosing an algo versus risk transfer is pretty important right now. We are best placed to help them achieve their execution goals, because we are the subject matter experts on our own tools.”

In addition to the liquidity issue, Mesick adds that customers also need to find a balance between minimising market impact, reducing market risk and certainty of execution – the algo execution Trilemma outlined by the BIS. “At one end of the spectrum is risk transfer, and at the other end, the client would basically be taking unlimited market risk,” he adds. “Customers have become comfortable with these trade-offs and part of that has worked into the muscle memory customers have developed from using specific algos, from specific providers, with specific sets of configurations.” Mesick notes, however, that since August and the yen move the market has seen a significant uptick in volatility and a shift in market dynamics following the US election and more recently following Trump’s Liberation Day. These new market dynamics are fundamentally what is causing customers to reflect on what matters most to them, he adds. “Our algos are designed to adapt to these changes, but how well the customer is able to balance these execution trade-offs is something that varies depending on market conditions,” Mesick says. “This is going to change how customers rank those three elements of the Trilemma and that may lead to behavioural changes.”

Liquidity curation

In addition, it is important to ensure algo performance against the backdrop of the increasingly complex FX liquidity landscape, which has resulted in significant enhancements the BARX Gator suite over the past two years, Mesick explains. He adds that part of this development has been to focus on increasing the effectiveness of the sub-components, because those are shared and put together in different ways within all of the Gator algos. “How do we passively place? How do we access non-visible liquidity? How do we utilise the franchise? All of these are sub-components, the building blocks of our algos: Gator adapt, which is our implementation shortfall algo; Gator float and our Gator Twap algos, the core passive algos in the suite,” Mesick says. “What we’ve seen is that if you build those well, the algos can adapt to market conditions. But if the markets are significantly more volatile and you’re using a Twap for an hour, which may have been a perfect setting just a year ago, in today’s markets that might no longer be the case. And so we are continually driven to monitor and enhance those features within the algos to ensure our algo suite remains cutting-edge.” 

When clients opt to use a particular algo strategy or product, they are also essentially delegating liquidity management to the algo provider or algo broker, adds Asif Razaq, Global Head of FX Algo Execution at BNP Paribas. He explains that this is essentially the result of clients typically having limited access to that market data, whereas the banks themselves have a wider data set and a better understanding of the market data and liquidity across the different venues. In addition, algo providers interact with these venues on a daily basis, which means they are much better positioned than a client would be in terms of determining which venues are better versus others, Razaq says. “Generally, the task of managing and dealing with fragmented liquidity falls upon the banks and their understanding of which venues are good and which venues are bad, as well as how to best interact with those venues across different currency pairs at different times a day,” he adds. 

“Liquidity management is an active part of our day-to-day job as an algo provider. We are constantly analysing the different venues that we interact with – where we currently have access to more than 15 different execution venues. But do we proactively trade on all of them? No, we would probably be very selective on how we interact with each of these venues and with what flavour algorithm we plug into those venues.”

Razaq continues: “The other aspect of understanding liquidity is knowing that when you connect to one venue, that is not just one liquidity pool. We have several pools within one venue and we would have these pools tailored per algorithm. Depending on the nature of the algorithm, we can work with those venues to ensure they curate a liquidity pool which is going to be suitable for a particular algo strategy. As a result, we have different flavours of liquidity with the same venue. This has been a significant area of focus for us since launching our algo service.” The FX market is also continuously evolving and so BNP Paribas is constantly reviewing and tweaking the pools and the number of venues that it might actively trade on, he adds. 

Fine-tuning the familiar

“This includes reviewing the liquidity curation for each algo and determining the right venues to provide good, low impact execution for our clients,” adds Razaq. “We then need to decide – how do we want to interact with a particular venue? How do we post our orders onto that venue? And which order type we utilise on each of those venues. For some venues, we may post limit orders which are lit, we may post some limit orders which are dark. Dark passive orders  are gaining popularity as the market is becoming more sensitive to orderbook changes where we are seeing market impact on the back of placing lit orders. We need to be very cautious of where and how we show interest, and analyse and monitor market impact on behalf of our clients. This is because clients do not have the access to that level of market data to do the analysis themselves. That is where we step in. Our algo clients delegate this analysis to us, where we tailor-make the algorithms and the order placement strategies to ensure that we are maximising our opportunities and sources of liquidity.”

In addition, many of the leading algo providers are now turning their focus from algo development to ensuring that existing strategies are instead tailored to specific market conditions. James McGuigan, Head of FX Algo

Product at Citi explains that this can be a result of some clients having a degree of ‘new algo fatigue’. This stems from the significant commitment needed from a client to understand what the main features and aims are of any new algo product when it is introduced, he explains. 

“It includes a learning curve around how best to use the algo and – most importantly – being confident that it will perform better than those they already have available,” adds McGuigan. “This requires a notable investment of resource and a non-trivial element of risk for clients. We see this most in the liquidity seeking/opportunistic style strategies as these inherently provide freedom for our quants to further exploit in the search for gains.” He continues: “The focus on optimising our existing Arrival and Peg strategies has led to significant performance improvements in execution quality, for example. We have been able to both improve our slippage metrics whilst also decreasing the amount of time taken to execute any given amount. This gives clients familiar tools that continue to return incrementally better performance.”

Transparency and insights

Laming also notes a general and ongoing market trend of algo volumes increasing across the board and for all client segments. He adds that as algo business continues to grow, it makes sense for the market to look at the FX Global Code and ensure the principals are being applied in the algo community. “At CACIB, we have always been very proactive around algo transparency,” Laming says. “We provide our clients with a lot of data around their algo executions and encourage them to use this data as part of their own analysis. For instance, we are already working with several third-party TCA providers and we are working on onboarding additional ones. The data is there on the sell side but very often the buyside lacks access. We believe that transparency and sharing these insights with clients is essential.”

According to Laming, the growth in third-party TCA is a good feature for the FX algo market as it adds an extra level of standardisation, which is a very good starting point for discussing algo performance with clients. “In addition, pre-trade discussions are always interesting for buyside clients,” he adds. “At CACIB, we believe it is important to strongly segregate those pre-trade discussions with our team from our trading activity. This allows our clients to come to ask us for analysis or insights to help inform their planned algo executions, which they know will not be used against them. We are more than happy to have this engagement with our clients, because in helping them make the best trading decision we are helping them to achieve the best outcome possible. We want to build partnerships with our clients and so if the outcomes from their algo execution is good, then we are happy.”

Laming adds that at CACIB, FX algos are almost tailor made for clients and the team is keen to tweak parameters to match the client expectations and needs again and again. The discussion with clients is the important point, as not every algo strategy will work for every type of client. “We have some corporates who could be risk averse, for instance, or some hedge funds which might have a lot of appetite for risk,” he adds. 

“So we do not try to offer the same product across the board, we often will tweak the parameters and adapt the curation of liquidity pools accordingly, but always to match the client’s expected execution outcomes. Our approach is that we start with the client and then we evolve and adapt the product to meet their needs. At CACIB, innovation is part of our DNA. We are always keen to add new features or products, but if there is little interest from the client then it would ultimately be unproductive. So for us, discussion and collaboration with clients is the heart of everything we do and which very often leads to a win-win situation for both us and our client. Our approach is clients first, then we adapt.” 

Finding an equilibrium

Volumes in the algo space are continuing to increase, which Kataria says goes hand-in-hand with a higher level of focus on best execution and transparency. 

“We are on the forefront of our clients minds because of our ability to meet those needs,” he says. “Customers have noticed that we are a top tier provider for algo execution across a myriad of metrics. We have fine-tuned and tailored the sub-components of all of our algos to create a best-in-class product. Which means they also come to use when faced with these liquidity challenges to help then figure out how to utilise our tools to find that balance between cost, efficiency and execution quality. Third-party analytics also demonstrate our performance, for example, our liquidity seeking algos look really, really good – and our clients recognise that from their own data as well.” 

Kataria continues: “We also put our money where our mouth is. Our spot desk uses our algos and our traders are the biggest active users of our algos here at Barclays.” From a sales perspective, the quant team, the statistical modelling team is often included in client calls and meetings, adds Kataria. 

“That is part of the increased transparency that we offer our clients as they now are more questions that are more in depth and more quantitative focused. Therefore, we try to bring those resources to them so we can have those higher level conversations and go deeper into answering these questions,” he says. “Our goal is to bring more certainty to a very uncertain market and to increase transparency and to bring increase efficiency in the spot market. Our investment in the quant research over the past year and a half is testament to this commitment.”

There is also a balance that needs to be found between internalisation and the various types of liquidity available in the market, whether that is lit pools, dark pegs or mid pools, Mesick says. “It is pretty clear that at the extreme ends of the spectrum, all dark or all lit, is not where this equilibrium will be found. Instead, being able to take a hybrid approach to liquidity and knowing which venues within those different types you should be in at any given time is really based on the amount of quant research that the business is doing.” 

Mesick adds that the quant team has been involved in a significant amount of research which the team is able to harness that within the models to be more dynamic about where and ultimately, what that does is it scales. 

“The markets aren’t going to consolidate tomorrow, there is always going to be the these ebbs and flows. The key thing is that you have the quantitative framework and teams in place to recalibrate over time. And secondly, that the algos have been built in a dynamic enough way so that they are not historically tuned but can adapt to current conditions, which is why our algorithms have performed exceptionally well over the last two months,” he says.