FN: How would you summarise the key benefits of using FX execution algos and what sort of questions would you expect to be asked by clients who are just starting out on their FX algo trading journeys?
FX algos offer a powerful, flexible execution tool that puts clients in control, tailored to hit their specific trading objectives. The key benefits offered by FX Algos include minimising market impact, increased transparency and customisable strategies, resulting in a lower execution costs for clients. For new clients in particular, understanding the implication of these components is essential to their overall trading journey. Firstly, what is the quality of the analytical tools from pre-trade, real-time and post-trade and will this provide increased transparency around the quality of the execution? How can this help develop specific metrics? What are the various different strategies available and how can they minimise transaction costs? What is the choice of external trading venue, and how significant is internalisation? It is worth noting a child order executed by the algo internally will benefit from curated liquidity which will reduce information leakage. The overall cost reduction is quantified by comparing algorithm performance versus risk transfer price (RTP) – the electronically-streamed price at the start of the execution. It is always important for our clients to have an in-depth analysis of the performance of each algo order.
FN: Price and speed are important factors in FX execution decisions but why is market impact also of concern for many buyside firms and how can algos help to reduce it?
FX algos offer a systematic, data-driven way to minimise the execution cost while achieving a client’s objectives where the price, speed and market impact are the key risk factors. Market impact can significantly negatively affect execution performance and therefore managing market impact is a critical success factor in algo value creation, leveraging on the depth and resilience of liquidity. Advanced algorithms mitigate directional flow leakage through adaptive slicing techniques with appropriate child order types. These algorithms leverage high-frequency microstructure information and diversified, curated liquidity channels with minimal interbank exposure. Quantitative models are employed to optimise the pace of execution by analysing observed market impact signatures, balancing the goals of speed and cost efficiency.
FN: How do leading providers go about computing the level of expected Market Impact that their algos will be able to deliver?
Leading providers typically engage large quantitative teams to study market microstructure, curate liquidity, design and implement execution algorithms. These teams continuously refine their market impact models through rigorous calibration against historical algo performance data, which account for market conditions and chosen parameters. The effectiveness of these models is validated using in-house analytics and third-party services. Clients receive transparent insights on expected performance metrics and associated trading risks.
FN: How are leading FX algo providers tackling the problem of differentiating their offerings in this increasingly competitive market? For example, by taking a different approach to liquidity management.
FX algo offerings are highly correlated to the quality of eFX market making infrastructure and client franchise components integrated into algo strategies and analytics. The flexibility of a proprietary platform managing the algo models and smart order router (SOR), enables banks to curate liquidity and continuously monitor performance. The FX market offers many liquidity styles: primary to secondary, fully disclosed to anonymous, firm to hybrid, and a new generation of discrete peer-to-peer liquidity. Providers implement them differently depending on their product strategy and IP. Internalisation is vital in FX algo execution, especially during periods of high volatility. Each bank’s approach is shaped by its unique combination of client profiles, franchise strength, and operational framework.
This determines capacity to leverage internal eFX market-making resources, creating a blend of liquidity options that set banks apart and benefits algo users.
As the eFX market rapidly evolves, adapting liquidity curation for efficient algo execution across new currency pairs offers a competitive edge. This is evident in recent developments to adapt existing algos to the electronification of NDFs and precious metals. To manage the market evolution, many banks offer a full complement of strategies: in HSBC’s case, it’s a suite of 8 algos that includes POV (percentage of volume) and the multi-currency basket, based on correlation optimisation.
FN: Why has internalisation become an important consideration for increasing numbers of FX execution algo users and what role can it play in helping to improve execution performance?
Internalisation has become a fundamental differentiating factor when executing FX algos, acting as an additional execution pathway with minimal information leakage. This has been particularly noticeable given the increased market volatility and complexities of the market microstructure. Markets have become increasingly sensitive to directional flow on public venues, which can impact market pricing and bid-ask spreads with the execution of larger orders. As FX markets remain primarily over-the-counter (OTC), the advantage of internalisation can be realised via algo interaction with a significantly large OTC liquidity pool, rather than just algo versus algo.
Major liquidity providers offer significant levels of internalisation due to the scale and diversity of their operations. However, achieving a high degree of internalisation itself is not the objective. The goal is to deliver better outcomes for clients, which principal liquidity supports through consistently flat markouts—demonstrating minimal market impact and low adverse selection.
RT: Why are pre trade analytical toolsets becoming increasingly important for some buyside algo trading firms?
Pre-trade analysis is becoming increasingly important as it allows the trader to bring post trade analytics of previous executions to the forefront of decision making. When a trade arrives on a trader’s blotter, they want to bring any learning from previous experiences to the forefront of deciding how to trade. The information must be delivered in a transparent, compliant and auditable way. The decision to trade may well be also driven by information pertaining to the underlying market conditions and any key events that may be about to be released.
FN: What work is being done by FX algo providers to enhance their Transaction Cost Analysis offerings to help clients make more informed decisions about algo selection and execution quality?
Technology, regulatory frameworks and margin reduction across the industry have impacted how clients quantify the quality of their execution. As a result, the data offered in transaction cost analysis and post-trade reporting constantly adapts to these new requirements. TCA gives clients the transparency and all the information required to optimise algo performance but this process extends beyond a simple post-trade report.
Recently, the focus has been on providing data in flexible formats since clients increasingly request API integration for independent data analysis. Third-party analytics providers have a growing role in the market. Accurate, resilient integration is vital for reliable service and performance comparisons. In addition, increasing focus on pre-trade performance stats underscores the need for accurate data.
FN: The delivery of algo analytics can be fragmented. What is being done to improve their usability?
There is considerable development to align the usability of FX algo analytics across multi-dealer trading platforms. Banks’ single dealer platforms, however, have some of the most comprehensive analytics tools in the market — from analysing the market landscape to flexibility, transparency and interactivity throughout their algo executions. In the case of HSBC, these types of tools have been designed within the HSBC AI Markets trading terminal and HSBC Evolve single dealer framework.
RT: As more asset managers and corporates start to adopt algo trading toolsets what would you like to see done to make disclosures easier to understand for clients of varying degrees of sophistication so that they can match their individual execution requirements with the most appropriate execution algo?
The industry has a large divergence in the understanding of disclosures. When confronted with a disclosure for the first time it may seem very daunting. The journey of improved execution is iterative in nature and understanding of things such as disclosures should also be viewed this way. When a trader has chosen a method to execute, he is aligned to a target outcome and should have a good understanding of what he is doing. At that point he should be thinking of improvements that he could make to his decision making at the point of trade and during the execution of the trade.
FN: How much demand in the future is there likely to be from buyside firms for more customised FX algo trading solutions? For example, to adapt an existing algo to make it a perfect match for their particular needs and requirements.
Buyside demand for customised FX algo solutions will continue to grow, fuelled by a need for a differentiated execution, better control and more integration with internal analytics and systems. This process has already been initiated given each buyside firm has its own infrastructure and execution strategy, whether seeking alpha or simply measuring execution costs.
Almost all algo elements can be customised — for example, HSBC offers a floating principal order algo solution by aiming to reduce market impact. Through configuration adjustments, the algorithm can change its mode of interaction with HSBC’s central risk book by placing pegged passive orders and matching them with incoming client orders.
RT: In what ways is the growing use of execution algos in FX likely to impact on the skillsets required of traders to manage them?
Over time the skillsets of traders have evolved. Traders now must be more skilled in data manipulation to ensure that output from data sets can be used in a way that is beneficial to the trading outcome. Not only must they ensure that data is clean and relevant, they must also ensure that the output is presented in a way that is easily understood to differing consumers. Looking forward, the ability to manipulate this data in an automated manner will be key. Programming skills such as Python will be essential. There is the hope that evolution in automation will put solutions at traders’ fingertips, allowing them to spend more time on market analytics. We are also on the cusp of great steps forward in artificial intelligence and machine learning, which will bring other opportunities and repercussions for traders to be mindful of.
RT: We are starting to see the arrival of guidelines and standards in algorithmic FX trading. What benefits do you think these will deliver particularly in the years ahead as the level of complexity of this style of trading continues to increase?
Guidelines and standards in algorithmic FX trading can only be a good thing. In essence, trading desks will have to ensure there is adequate governance structure around the algos they use and how they use them. The level of complexity is already growing, and traders need to make sure that they understand the nuances of the algos they trade and how they are constructed.
Transparency and governance are something that supports the principals of the FX Global Code. With that in mind, traders will have to fully understand how algos work. Better understanding of urgency, internalisation, nature of order posting and differing natures of pools of liquidity are a few of the parameters that traders will need to better understand.
FN: How far are next generation technologies like AI likely to be applied in algorithmic FX trading and what can be done to ensure that more sophisticated techniques like Machine Learning can be governed sufficiently, particularly as some practitioners believe that ML could create new market fairness and stability threats that may require new distinct governance frameworks?
Algorithmic execution already uses machine learning techniques, and appropriate governance on probabilistic outcomes is in place. Execution algorithms typically have well-defined goals and constraints, and the development process is thoroughly verified and highly regulated.
We expect to see generative AI being used in the market in the near future to enhance and simplify client interactivity with an algo’s API, as well as to aid in pre-trade decision-making and performance interpretation.
FN: How much potential is there for taking a more multi-asset approach to FX algo development, particularly where FX is not sitting alone but is part of other trades being undertaken by buyside firms?
Equity and fixed income managers have been generally less proactive at managing their FX risk, resulting in fragmented FX execution behaviours. Recent market dynamics have, however, prompted global investors to quantify the FX returns in their underlying investments. The HSBC FX Basket Algo caters to this type of buyside requirement, and has been designed to optimise currency basket execution through FX correlation analysis. This new algo strategy is suitable for any cross-asset portfolio manager looking for an efficient FX transaction, as defined by a reduction in electronic execution costs and operational workloads.
FN: As buyside firms seek to deepen their understanding of FX algos, how far should sellside providers be willing to go in offering more comprehensive execution coverage specifically focused on algorithmic trading?
The algo analytics suite equips clients with dynamic tools to help them achieve their execution goals. This starts with customised onboarding, for each client, product education and continuing throughout the algo order’s lifecycle. Comprehensive in-flight coverage is available via segregated algo teams or self-serve analytics tools, or a combination of both depending on each clients’ focus. The service continues post-trade, with detailed scrutiny of performance to help clients assess whether their objectives were met, whether alternative strategies should be used, and, in some cases, whether bespoke configurations should be incorporated.
RT: As the use of algorithmic trading in our industry gathers pace what implications does that have for the nature of the trading room of the future?
The trading room of the future will have more reliance on automation and data analytics. Manual tasks will be moved away from traders’ fingertips. Smaller trades which don’t need a trader’s attention will be auto traded with a governance structure to capture any exceptions for analysis. This will allow traders to spend more time on trades that require greater focus, to generate as good an outcome as possible.
There will be more reliance on the delivery of data analytics at point of trade, to help the trader optimise their outcome. Evolutions in artificial intelligence and machine learning are on their way, so desks are making preparations to take full advantage.
