Why the case for using FX execution algos will strengthen in 2026

According to Dr John Quayle, Head of Client Algo Execution at NatWest Markets, market fragmentation is set to prove the biggest driver in increasing client interest in FX algos this year. He explains that while there should be a move towards greater consolidation among FX venues, in reality quite the opposite is happening. For example, where electronic markets are nascent, such as forwards and swaps, Quayle notes that there has actually been a significant increase in the number of venues available over the past two years, with more expected for 2026. Even for spot, which is by far the most mature FX market, novel venues such as One Chronos will be arriving shortly, Quayle adds. 

He stresses that making full use of these multiple venues simultaneously can only be successfully achieved by something like an algo. “In forwards, we will soon see the first client facing algos which will aim to maximise the use of the available liquidity across venues and, perhaps more importantly, leverage the ability to place resting orders,” Quayle explains. “These are the ingredients which allow a spot FX algo to, on average, reduce transaction costs relative to a risk-transfer trade and I have every confidence that algos in the forwards and swaps space will be able to deliver the same benefits to users on the buy side as well as in banks.”

In turn, Quayle notes that the candidates for product types other than spot which could show a significant uptick in the development and use of algos are not only forwards and swaps but also NDFs. However, he argues that although NDF algos have been around for a while, as have various NDF venues, the actual uptake of NDF algos has been disappointing. “An algo market survey that we at NatWest conducted in 2025 showed that 75% of respondents do not use them, and that proportion was unlikely to change. Commonly used platforms also report little update,” says Quayle. On the other hand, he believes that the picture for algo growth in the forwards space will prove to be very different, with some algo activity already being reported on existing venues. “With the number of venues set to grow even further this year, more banks will routinely start streaming swaps. This combination is expected to create enough critical mass to make algo execution a viable possibility in forwards as well.”

Potential for NDF growth?

The case for FX execution algos in general will also continue to strengthen into 2026, as buy-side trading desks face growing complexity, scale and performance expectations, says Oleg Shevelenko, FXGO Product Manager at Bloomberg.

He explains that because FX trading is increasingly multi-asset, multi-instrument and global in nature, this in turn is making traditional voice-based execution harder to scale, further accelerating the move towards electronic and automated workflows. “Algorithmic execution offers consistency, repeatability and stronger governance, which are becoming essential rather than optional,” says Shevelenko. “At the same time, expanding algo coverage beyond spot FX into NDFs, swaps and precious metals means that firms can standardise execution workflows across a broader portion of their FX activity. Strong adoption trends – FXGO, for example, has seen over 40% year-to-date growth in algorithmic trading volumes spanning both spot FX and NDFs – demonstrate that both existing users and new entrants are increasingly confident in algos as a core execution tool.”

Shevelenko notes that algos also help buyside firms navigate market fragmentation by intelligently routing orders, managing partial fills and optimising execution across multiple liquidity sources, which reduces reliance on any single counterparty or venue. “Meanwhile, algorithmic execution is easier to embed within structured governance frameworks, providing audit trails, execution consistency and clear performance measurement, enabling firms to meet regulatory and governance pressures,” he adds. “Technological advances such as improved analytics, pre-trade optimisation tools and better integration of pricing and workflow tools also allow algos to deliver measurable execution benefits.”

In contrast to Quayle’s observations, Shevelenko believes that the most significant traction beyond spot FX is still expected in the NDF market. He argues that NDF’s actually appear to be following a similar evolutionary path to spot FX, where increased electronification and algorithmic execution are driving greater efficiency, pricing consistency and standardisation. Improved regulatory clarity and broader access to electronic liquidity are key factors supporting increased institutional participation, according to Shevelenko, who adds that although NDFs have traditionally been used primarily for hedging, “speculative activity is now on the rise, reflecting a broader and more sophisticated participant base.” Alongside NDFs, Shevelenko agrees that algorithmic execution in FX swaps and precious metals is also gaining traction as platforms expand instrument coverage and buy-side firms seek to streamline execution across multiple FX products. 

He adds that asset managers and hedge funds are the client segments which are expected to increase their use of FX execution algos at the fastest pace over the next year. Shevelenko explains that asset managers benefit from algos through scalable, repeatable execution across portfolios, stronger governance and enhanced analytics to support best-execution requirements. On the other hand, hedge funds, particularly macro and systematic, are drawn to the speed, consistency and ease of integration that algorithmic execution offers, particularly as participation in instruments such as NDFs grows, he adds. 

Data speed and security

“FXGO is focused on delivering tailored roadmaps for these clients to meet these particular demands, enabling them to do more with less and ultimately process more orders,” Shevelenko says. “For example, FXGO offers pre-trade order netting and optimisation tools to reduce execution costs and streamline trading; basket algorithmic trading functionality; bulk-routing capabilities to both resting and algorithmic orders, primarily used by asset managers and the ability to pre-create algorithmic orders with specified parameters that can later be automatically routed via FXGO – typically utilised by hedge fund clients to optimise their workflows.”

For the buy side, the transparency offered by algo execution is a key driver behind the increased interest in their use, according to Richard Turner, Senior Trader, Currency Solutions at Insight Investment. He explains that greater transparency allows algo users to interrogate the data that is supplied during or after execution, such as the TCA provided in-flight or post-trade. Turner adds that this not only allows the user to satisfy their regulatory obligations, but also enables detailed analysis at a child order level, ie decay/impact etc. “The user can also assess the differences in the outcomes from using different pools of liquidity and appropriately channel their flow to the pools that satisfy their requirements,” he says. “Technological advances that we are particularly focused on this year in terms of FX execution heavily relate to the speed of processing data and the security of our data.”

Turner adds that potentially the firm would be interested in being able to utilise the speed associated with artificial intelligence to enhance the evaluation of the outcomes of previous trades, compare current market conditions and enable informed choices when executing algos. “All of this needs to be achieved, however, without putting any of our data or infrastructure at risk to outside forces,” he warns. Another key trend this year is again expected to be greater traction in the development of algos for FX swaps, but Turner also predicts that FX options algos could prove to be a significant addition. 

Algos vs risk transfer

Turner notes that in terms of FX swaps, when executing an algo the forward points may change over the period of execution. The ability to roll forward child orders as they are executed would then prove to be optimal, from both a risk mitigation and ease of execution standpoint, he adds. “Both algo provider and user would benefit from transparent and timely execution as well as more efficient price discovery, but participants would have to get more comfortable with the automated forward roll, especially when trading in size,” he explains. 

Option algos are also already widely traded in equities and, according to Turner, it is “just a matter of time” before they are more widely used in FX. However, he advises that, once again, participants will need to get comfortable trading electronically but eventually, when transparency and price discovery are satisfied, FX options will certainly start to be traded more widely electronically. “One additional caveat is the ability for providers and users to embrace technology from a resource perspective,” says Turner. 

Additionally, the further adoption of the FX Global Code should also encourage more use of algos as participants increasingly strive for transparency over outcomes, Turner predicts. On the other hand, he warns that one area which may challenge the uptake of algos is the competitive availability of good risk transfer prices that are currently on offer. He adds: “Users need to constantly assess the optimal outcome, weighing up taking timing risk on an algo versus removing risk but paying spread. These outcomes will ultimately determine whether more or less use is made of algos by the buyside long term.”

The ongoing data challenge 

Quayle agrees, adding that one of the key difficulties that may be holding some buyside firms back from greater adoption of algos generally is the challenge of how to more accurately benchmark algo performance. This is because when clients are executing spot on an RFQ, or via stream, the user can see upfront which price of several is the best, tracking hit rates over time to fairly quickly build up a picture of the relative strengths of the LPs being used, he explains. With an algo, however, Quayle warns that the best measure of performance (such as which algo has least market impact) is the slippage vs inception mid, ie the difference between the rate that the algo achieves and where spot was when the algo was initiated. “Here, you only get one data point per algo, so it takes time for even the most active algo users to build up enough data to say which algo is better than any other. This is a major hindrance to buyside firms who are seeking to gain more confidence around their execution choices when it comes to algos,” he says.

As a result, Quayle believes that greater use of third-party analysis, which combines anonymised data from multiple algo users, is key. Crucially, this needs to be done in a manner which gives both buy side and LP participants confidence that the data is reliable, he adds. “However, gaps in this data continue to persist, as do questions over its integrity,” says Quayle. “Therefore, taking steps to increase the quality of available data would be greatly beneficial to the market as a whole.”

When looking ahead at the relative maturity of the FX algo space versus the potential for further growth, he notes that in many ways the algo products themselves can already be said to be quite mature. “This is because algos have been refined extensively over the years and now offer increased performance and flexibility”, according to Quayle. He adds that algo strategies can now often be tailored for varied use cases, while also efficiently bring together liquidity from all parts of a highly fragmented market. 

Adoption rates by region

However, whether the market itself can now be said to be mature is harder to say, Quayle explains. He argues that algos are still regarded as ‘exotic’ by many market participants and, although volumes continue to grow, they still remain tiny in comparison to other methods of spot execution. “There are also challenges around benchmarking and performance measurement and the market has not yet found a way to provide high quality and transparent data on this. The fact that many buyside participants still allocate algo volumes based on spot or RFQ hit rates, rather than how good the actual algo is, tells us that there is some way to go to achieve a place that might be called ‘mature’,” Quayle notes.

Innovation in the development of algo strategies is also key to attracting new users, with the market having moved beyond offering only basic execution strategies towards more sophisticated, outcome-driven algos, Shevelenko says. “At Bloomberg, our approach to innovation is focused on deepening collaboration with algo providers and onboarding new partners to broaden instrument coverage, especially in emerging markets and derivatives,” he adds. “We also integrate composite pricing, real-time news, analytics and cost-modelling within clients’ trading workflows for improved execution quality and operational efficiency.” Shevelenko shares some examples of recent enhancements, including basket trading workflows; multi-step execution for forward and swap orders, enabling clients to execute the spot component via an algo or a benchmark order, subsequently working with the sales desk to roll the trade to the desired settlement date and improving workflows for capturing comprehensive algo event data to support enhanced performance analytics.

In terms of regional differences in algo adoption rates, Shevelenko adds that adoption in APAC continues to outpace other regions, which he believes is due to the combination of strong growth in electronic FX trading and the region’s established central role in NDF markets. “That said, while adoption is accelerating, particularly in emerging markets, electronification and algorithmic maturity remain uneven,” he explains. “However, NDFs which are continuing to develop, can see fragmented liquidity and less mature strategies than in spot FX, leading to varied adoption rates across regions and currency pairs.” 

Adoption of AI and ML

According to Shevelenko, there are also different merits to be found between using in-house and third-party algos, with in-house algos offering greater customisation and control, while third-party algos provide speed to market and proven execution logic. Ultimately, he argues that access to deep, diverse liquidity is “critical to performance”. He continues: “FXGO offers both direct access to liquidity via our platform UIs and APIs, as well as a comprehensive suite of dealer algos, allowing firms to choose the model that best fits their needs.”

Improved market structure and deeper liquidity access is also key to the acceleration of FX execution algos, Shevelenko explains. He notes that growth in regulated venues, streaming liquidity and NDF-specific algorithms serves to increase confidence in electronic execution for a broader range of participants. “From a platform perspective, multi-dealer access and strong pre- and post-trade transparency within a single regulated workflow, as well as continued innovation, all play an important role in driving the industry’s adoption of FX execution algos forward,” says Shevelenko. “Overall, the FX algo market remains on a strong growth trajectory, supported by structural changes in market structure, regulation and buy-side operating models.” He adds that algorithmic trading in spot FX is relatively mature, whereas algorithmic execution in NDFs is still at an earlier stage, characterised by limited trading venues, uneven liquidity across currency pairs and less mature strategies compared to spot FX. “Continued development of regulated venues, streaming liquidity and NDF-specific algorithms is expected to be a key driver of future market growth,” adds Shevelenko.

On the innovation front, Turner explains that a key change in the FX market regarding the adoption of AI and machine learning has been in the continuous improvements being made to FX algos. “Algo providers now use AI and ML to consume the multitudes of trading data that is available to constantly try to improve execution outcomes,” he adds. “This has led to a greater depth in liquidity achieved by market correlations being a tool for the sell-side algo developers to incorporate into their algo hedging strategies.” Turner notes that some algo consumers are also continuing to develop pre, intra and post-trade analytics which are more akin to rules-based decision making, as opposed to developing a complete AI and ML offering, which again would require a heavy focus on governance and supporting infrastructure.

Growth through education 

According to Turner, the buyside is also not focused on developing in-house algos, which he argues would be very resource heavy to run and maintain. “This is not just talking about the resource required to ‘run’ the algos, but also a reflection on the resource needed to govern and regulate them,” he adds. In addition, Turner explains that the additional challenge of AI and ML is the ongoing effort to maintain and improve outcomes, which he believes will only serve to further compound this resource cost. “Third party algos instead allow us to challenge the providers to improve outcomes by competitive innovation as well as furthering the in-flight experience for the user,” he says. Turner adds that the industry should also continue in its efforts to better demonstrate improved outcomes when using algos and in educating end users about the benefits of algo execution versus voice execution, which is often manual. “While users should continually challenge algo execution versus risk transfer (voice or electronic), they also need to understand the role algo execution can play as a part of the toolkit to achieve optimum and transparent outcomes,” he says. “When using algos, you have increased visibility over your trade alongside immediate options for how you execute (such as limits, aggression, LP selection etc).”

Ultimately, the algorithmic FX trading market is as mature as the challenge it faces, Turner explains. For example, he suggests that FX market structure could be greatly improved by having more focus on leg execution on ‘cross’ trades, such as better analysis of EUR/USD, GBP/USD and EUR/GBP fills in EUR/GBP execution. He adds that work also needs to be done to “further analyse the benefits of human interaction with the algo.” There is also scope for further growth in terms of developing algo execution alongside the objectives of the FX Global Code, Turner believes, including looking at how FX algos are expected to evolve with regards to options, swap, forwards and NDF algos. “The first hurdle, however, is educating existing and new users about the benefits of algo execution, notwithstanding the ability to opt for risk transfer trades when needed,” he concludes.