Exploring tailored FX execution algo trading solutions

There is no one-size-fits-all when it comes to algo execution, due to different execution objectives and target benchmarks among clients, explains Neil McClements, Head of EMEA Multi-Asset e-Sales, RBC Capital Markets (RBCCM). He notes that by partnering with clients, the bank is able to provide more tailored, optimised solutions. “We have clients who use our intelligent hybrid IS algo strategy, for example, but what a hedge fund determines as aggressive urgency can be very different to what an asset manager determines as such,” McClements adds. “Therefore we have multiple variations of our IS algo which are optimised for individual clients.”

Liquidity is also critical and there are now a myriad of liquidity sources available, each with its own nuances: lit markets (eg primary vs. secondary, full amount v sweepable, firm vs. last look etc), as well as dark, midpoint and internal pools, according to McClements. He explains that correctly harnessing these diverse liquidity types is fundamental to the success of algo execution. “Our algo suite uses advanced policy iteration techniques to achieve best execution by intelligently blending multiple liquidity sources in real-time, based on evolving market conditions,” McClements says. “We have also designed and built our algo architecture from the ground up, which provides us with much greater control and the flexibility to provide tailored, optimised solutions. We can easily customise curated liquidity pools, execution urgency, participation (POV), risk checks and many more parameters.”

According to McClements, client preferences for algo interaction can range from fire-and-forget through to constant calibration. By recognising this spectrum, RBCCM can offer intelligent amend capabilities, enabling clients to alter strategies or recalibrate execution parameters mid-flight, all without affecting their overall execution quality. He adds: “In addition, clients can utilise our ‘I Would’ facility, adjustable in real-time, to opportunistically utilise beneficial market conditions.” There are also clear benefits from adopting a hybrid approach which allows clients to capture the strengths of multiple execution styles in a single algo, McClements argues. “For instance, our BlockSeek and Scaling features, allow clients to incorporate IS features into all our schedule algos (TWAP, VWAP, and POV strategies),” he says. “Our LimitPlus strategy also offers clients the opportunity to scale lit quantity utilising POV, while also using BlockSeek to target duration. We can create tailored strategies to meet multiple client objectives, regardless of which overriding algo the client wants to execute.”

Need for tailored solutions

Matthew Fitzpatrick, Head of e-FX Pricing and Flow Trading Models, Financial Markets at Westpac notes that ultimately clients are increasingly seeking customised FX execution algo toolsets and strategies for several reasons, including specific trading requirements, the impact of market volatility and an appetite among some algo clients for enhanced levels of control. For example, he explains that clients may need to trade certain currency pairs at specific times of the day when liquidity conditions might be challenging and customised algos can help navigate these conditions more effectively. “In volatile markets, participants may also need the ability to adjust their strategies quickly,” he adds. “Having tools that allow for rapid adjustments is paramount to maintaining effective trading strategies. In addition, customised algos can provide clients with greater control over their trading strategies, allowing them to tailor their approach to their unique needs and market conditions. By addressing these factors, customised FX execution algos can help clients achieve better trading outcomes and manage risks more effectively.”

Furthermore, liquidity is foundational to the customisation and construction of FX execution algos, as it influences strategy design, adaptive behaviour and risk management, Fitzpatrick adds. “It also ensures that algos can adjust to real-time market conditions and mitigate potential market impact and slippage,” he says. “In addition, providers are now offering more modular algo trading frameworks by allowing clients to customise execution urgency, participation rate, aggression level and venue preferences. This flexibility helps our clients to build their own tailored and highly effective trading strategies.”

It is also important to understand why clients are knocking at the door of algo providers asking them to provide bespoke elements or customisations to their chosen algo strategies, says Patrick Guevel, Head of FX Algo Execution at Societe Generale. He explains that some of these requirements can be straight forward, such as removing a certain venue from the curated liquidity pool. “Then other clients ask us to share ideas around how the algos might be made even better for them and we might have already been evaluating making certain changes which we can then put on the table to create these customisations for that specific customer. However, when we create a specific change like this, we often will do this with the aim of extending it to wider pool of algo clients as well. If a customisation will be of wider benefit than to the single customer with whom we interacted to get this first, we tend to think larger and decide if the time and resources we have invested in this development could be leveraged by extending or enhancing the service as a whole.”

Customisation and construction

Guevel notes that sometime clients might ask for individual adjustments to the curated liquidity pool, which the bank is able to accommodate. “Ultimately, it is the client’s decision and we will respect that, provided it is within the framework of the execution. Additionally, we highlight that it might not be optimal to remove that source of liquidity,” he says. “In the algo platform, we take inspiration from the equity market to track live measurement of the market volumes which we compare with historical values, and we adapt our liquidity pool accordingly. That is one of the major enhancements we have made this year for our FX algo suite.” The benefit of this new enhancement is that when there is a change in trading, for example on a bank holiday, then the tool will automatically detect that there is reduced activity and will adjust the parameters of the algo accordingly, Guevel explains. He adds: “This is all completely dynamic and was the direct result of discussions we had with our customers on how to best to manage these kind of changes in unusual market liquidity.”

In turn, as algorithmic trading becomes increasingly prevalent, State Street clients have also become more knowledgeable and are increasingly looking for FX algos that can execute according to their preferred nuances, adds Daniel MacGregor, Global Head of e-FX Platform Sales at State Street Markets. He adds that algo clients require the ability to customise venues and liquidity pools, execution styles and speeds, participation of volume parameters, while also accessing customised strategies with bespoke logic to help them achieve different objectives. “However, as more customisations becomes available, liquidity providers still need to build the customisation parameters to meet the desired outcomes and objectives of the client base,” MacGregor says. “At times, clients select a particular type of algorithm simply because it is widely supported by brokers, even if it is not designed to meet their specific goals. This can result in outcomes that fall short of expectations.” He explains that while a client might appear to be focused on achieving a certain benchmark, their true priority may be minimising the market impact of their trade within a set timeframe. In this case, their objective shifts to reducing the standard deviation from the benchmark rather than just beating it, MacGregor explains. “Such unique requirements drive the development of customised algorithms tailored to address these precise needs,” he adds.

It is also proving crucial for providers to have modular building blocks or parameterised features so they do not have to rebuild bespoke algos from scratch, MacGregor warns. “State Street algos are designed with flexible architecture for quick, easy customisation to meet clients’ trading needs,” he adds. “For example, our TWAP and VWAP algos use similar building blocks run different underlying curves. The same type of “schedule” algos are easily customisable to follow different curves generated by an analytics module tailored for a client’s trading objective. Similarly, by rolling out our version of the ‘franchise skew algo’ within our existing FLOAT algo, we decreased time to production by not building a standalone algo and we have increased our internalisation ratio by 50 percent in 2025.”

Collaborative approach

In equities, the co-development of bespoke algo toolsets between in-house quant teams and clients is already a common concept and it is something that RBCCM also actively supports in FX, adds McClements. Clients benefit by having direct access to an outsourced quant and development team aligned on the same execution objectives, which according to McClements can then lead on to the client achieving a superior outcome. “Partnering with clients to build bespoke solutions to fulfil their specific execution requirements ensures a higher value proposition, and helps provide transparency which is critical for internal stakeholders to understand how the algo executes and why,” he adds. In addition, RBC algos are also primarily designed to adapt to liquidity conditions in real-time, which McClements says helps avoid the risk of overfitting to historical data. “For example, we have developed a unique Limit Order Model (patent pending), derived from a solution to the Multi-Armed-Bandit problem, which is able to intelligently overlay order urgency and size over the liquidity space to help achieve the best execution possible, while still maintaining a low footprint and minimising information leakage,” says McClements. “We are also able to partner with independent third-party TCA providers to provide clients the tools necessary to analyse execution quality. RBC has always been at the forefront of thought leadership and transparency on FX algo execution and has partnered with the world’s largest clients to provide innovative solutions. We provide a global real-time algo execution advisory service, and partner with leading independent TCA providers including both BestX and TradeFeedr.”

Real-time control of algos also allow for a high level of customisation, including adjusting the level of aggression or participation, changing the composition of liquidity providers, pausing and resuming the algo and changing the strategy mid-flight, says Emily Gaedtke, Head of Electronic Execution & Pricing Portfolio Management, Financial Markets, Westpac. “In addition, adaptive algos are being offered to blend execution styles in a single strategy. These algos can take some of the decision work out by allowing customers to define how they want the algo to behave depending on market conditions, which is particularly useful in fast-moving markets,” she adds. Furthermore, while co-development of bespoke FX algo toolsets is not common practice according to Gaedtke, it is possible that buyside firms with specific and unique trading requirements might consider this more integrated and individual approach. “Providers can also personalise algo strategies by analysing customer dealing behaviour, including price sensitivity and market impact – and then suggesting strategies based on this information,” she says. “Providers are now increasingly offering varying levels of value-added support services such as real-time TCA, chats with algo specialists, digital assistants, and AI-generated insights to further improve the algo trading experience for clients.”

Demand for modular algo trading frameworks with customised components has also increased in recent years, adds Guevel. “Only around five years ago FX algo offerings were far more rigid, but then we started to decide that the customer can change everything but the currency pair and the side of the order when executing,” he says. “For the rest of the parameters, we made it completely flexible. One area which we thought would prove to be of great importance was the ability to change the algo style in flight – so a client might start the passive algo, only to find that the liquidity is far from optimal. One of our customers was wondering if we could do something to solve this issue, which is why we created the capability to switch to another strategy, such as Falcon.”

Achieving successful outcomes

Another area which SocGen developed was the ability to change the end time of the algo execution during its lifespan. “For example, our passive algo ends at the same default time, but a client may want it to finish earlier than that. We enabled this live parameter change in the algo and it is now relatively simple to modify”, Guevel says. ”One important thing to consider as well is that the market is at a point where there is a race-to-zero on fees, which makes some providers reluctant to invest in changes unless there is a compelling reason to do so.”

Mary Leung, Global Head of Client Algos at State Street, adds that at State Street, the belief is that too many algorithmic strategies within an offering can overly complicate buyside algo usage and be counter productive. Instead, she argues that a streamlined algo offering with additional flexible parameters is the best way to provide a toolset that can blend styles. “For example, incorporating an Implementation Shortfall (IS) or Percent of Volume (POV) parameter within our flagship ‘FLOAT’ strategy, rather than having separate IS and POV strategies, allows clients to engage or change the algo on the fly within the same “FLOAT” strategy,” she says. “Additionally, we have deployed a bespoke ‘algo of algo’ strategy, providing the flexibility to run all FX algo strategies as individual strategies under one parent notional.” In turn, Leung notes that the collaborative approach of in-house quant teams working directly with clients to build more customised solutions is an approach which should be the foundation for building all algorithms. “Whether it’s solving workflow problems, providing greater transparency in Smart order Router (SOR) execution, enabling bespoke liquidity construction, or developing innovative algos with unique features, clients have always been at the heart of State Street’s algo offering. We have developed multiple FX algos and enhancements based on clients’ objectives, thoughts and ideas and we look forward to continuing our partnership with clients as the market evolves,” she adds.

However, Leung warns that having a ‘one-size-fits-all’ algo will never work, which is why flexibility in parameters or ‘backend customisation’ is paramount for success. “As an example, some clients may think a standard ten percent POV rate is slow, while other clients think this is too fast,” she says. “Some clients want large order peer-to-peer matching, while others only want smaller peer-to-peer clips. Having this customisation available is crucial to fit the expected outcomes of various clients. Additionally, having the ability to customise algo liquidity pools by client/user leads to more successful outcomes for both the clients and the LPs,” Leung concludes.