Technology has changed the way markets function. While trading technology has pushed the limits of what traders can achieve in the markets, regulations have also pushed them towards greater electronification. Algorithms are perhaps the most obvious example of the market’s shift towards technological solutions.
Despite their prevalence in equities, algorithms are relatively scarce in FX. The market’s fragmented nature and complexity have traditionally not lent itself to algorithm adoption. However, market practices are quickly changing as a variety of forces are driving firms towards increasing technological sophistication.
Stephen Bruel, Head of Derivatives and FX, Market Structure and Technology at Coalition Greenwich is well aware of the changes that are taking place. says “Algos are a vital part of the FX market structure, and will increase in importance as algo use in FX becomes more common,” he says. “The pace of change has remained somewhat consistent over time, but adaptation will continue.”
Understanding the drivers of change is essential if we wish to decipher the market’s attitude towards algo usage.
The current state of algo FX Execution
The use of algorithms in FX has always been a persistent topic. In their recent Market Structure & Trading Technology Study, Coalition Greenwich reported that just 17% of FX traders surveyed utilized algorithms to execute trades. Furthermore, 25% of FX trades are reportedly executed via an algorithm. These numbers are extremely low compared to equities, but taking a closer look reveals interesting patterns.
Algorithm use is prevalent in the FX spot business where trades lend themselves to algorithm use very well. Other lines of business such as FX options and NDF’s witness lower to no use, and the reasons are easy to understand.
The nature of these trades makes them difficult to standardize via an algorithm. There are both practical and technical barriers to greater adoption. Trade terms, such as fixing, are easier to define via traditional voice channels. On the technical side, NDF algorithms might be incompatible with existing OMS platforms. Lastly, NDF trade volumes currently do not justify increased investment.
Low algorithmic usage is also explained by their relatively late arrival in the market. Compared to traditional voice and electronic execution methods, algorithms are still unfamiliar to traders. Aside from inertia, voice trading lends itself well to relationship-building, something that is essential for buy-side firms seeking market colour before placing trades.
It must be noted that voice trading doesn’t directly compete with algorithm usage. As Bruel notes, “Market colour matters. Relationships matter. Therefore voice and algos are both integral to FX execution and need to sit side by side; traders want and need the flexibility to implement a trade execution strategy based on market conditions and other factors such as urgency, type of trade, and size of trade.” He points out that those factors will help determine which approach (algo, voice, or other) is appropriate.
Embracing algorithms doesn’t reduce the risk of FX execution, and technology offers firms solutions to measure risk. However, they tend to be complicated because FX transaction cost analysis (TCA) is a complex task.
Before adopting a new execution method, firms must have a clear picture of the risks they’re assuming. Existing TCA tools don’t simplify the complexity of measuring FX execution costs and their implications. Coalition Greenwich reports that TCA use is around 30%, with half of financial firms surveyed using it, and 20% of corporates following suit.
The lack of a central tape to compare transactions against is just one unique FX condition complicating TCA and hindering greater buy-side adoption. TCA and algorithm usage are joined at the hip, and limiting factors in both spheres are placing a hard ceiling on adoption.
Thus, despite the demand for greater transparency in execution and data usage in process management and analysis, algorithm usage is not yet witnessing equivalent adoption.
How algorithms are currently perceived
Despite the less than enthusiastic way FX has embraced algorithms, there is tremendous potential for growth. The FX market is heavily fragmented, witnessing multiple execution platforms, trade strategies, and time frames. Under certain conditions, algorithm usage presents a significant upgrade over current workflows.
“Algos are most popular when working a large order over time.” says Bruel. “But it is not just about market conditions,” he continues. “Algos are currently more commonly used for FX spot than other types of FX transactions, for example.”
While liquidity in the G10 currencies is very high, trading strategies change the picture considerably. Coalition Greenwich’s study sheds interesting light on the various conditions that market participants encounter when executing their strategies.
75% of traders surveyed reported that algorithm usage makes the most sense when executing a large order over time, with just 8% indicating they would not use an algorithm in this situation. Algorithm usage when executing small orders is almost universally not preferred, with just 8% of respondents indicating they would be inclined to do so.
Volatility in the markets usually pushes traders away from algorithms, since traders prefer greater control in such situations. However, other algorithm use cases such as large orders that need quick execution, orders in stable market conditions, illiquid pair orders, and multiple order execution, present less consensus amongst traders. There is great potential for algorithm adoption in these areas.
Technology invariably improves over time and as algorithms evolve, they will cater to more market conditions. In turn, this will prompt algorithm service providers to offer more options. Currently, Coalition Greenwich reports that liquidity-seeking algorithms are the most popular, used by 69% of respondents.
Time-weighted average price and volume-weighted average price occupy the second and third spots. Roughly half of the algorithmic order flow is liquidity-seeking. While technological improvement offers growth potential, recent changes to the FX landscape are pushing market participants towards greater algorithmic adoption right now.
Algo adoption growth drivers
FX execution quality has become a hot-button issue for corporates and the buy-side over the past few years. Recent changes to the GFXC’s Global Code have reacted to this demand and focused on defining disclosure templates that help traders understand execution algorithms and TCA.
Beyond disclosure requirements, the Global Code also clarifies fees, routine, and other components inherent to algo trading. These moves have come at the right time since the buy-side is keen on exploring the benefits of algo trading and TCA for overall execution efficiency. The problem is comparing and contrasting different solutions offered by service providers.
Bruel notes these changes with optimism. “The GFXC is approaching TCA similar to how they’re approaching algos – disclosure and transparency to provide more information to their clients,” he says. “This will help the buy side better understand how their trades are being executed when using an algo.” When asked how this might affect buy-side algo use he says, “This should serve to give the buy side more comfort and confidence when using an algo.”
The GFXC’s standards will remove adoption barriers and normalize the use of algo technology in FX execution. Given the strong connection between TCA and algo usage, there’s no doubt that both functions will benefit from the other’s adoption.
Thanks to these moves, the outlook for algo use is positive, even if the current picture doesn’t bear such observation. Coalition Greenwich’s study reveals that 69% of respondents believe algo usage will increase, with no respondent believing it will decrease.
Source: Coalition Greenwich 2021 Market Structure & Trading Technology Study
While corporate algo usage has lagged that of other market participants, increasingly sophisticated corporate desks are searching for more efficiency. Technology, via TCA and algos, offers clear solutions and increased adoption is imminent.
New asset classes are another area that could potentially trigger growth. Currently, the spot markets witness significant algo usage, but the potential in the NDF markets is huge. While existing volumes are low, due to the reasons previously mentioned, market participants anticipate increased algo usage in NDFs.
Coalition Greenwich’s study revealed that 12% of participants expect greater algo use with NDFs. While this isn’t indicative of huge growth, it points to a significant increase from current execution levels. One reason for increased anticipation is the evolving structure of the NDF markets.
UMR has changed trading economics, pushing firms to examine trading costs. While these changes have made greater TCA usage imminent, they’ve also transformed the NDF market. Clearing is expected to increase in NDFs which will lead to standardization.
Greater standardization will prompt more algo use. Coupled with the rise in TCA, this growth will only increase. Thus, while algo use in NDFs might not match spot usage anytime soon, market participants can expect significant increases in usage.
The link between TCA and algo use is critical to note, Bruel points out. “TCA and algos are distinct in many ways. Increasingly the buy side wants to measure the quality of their trade execution regardless of how they execute, but certainly when using an algo.” As technology improves, enhanced TCA will undoubtedly fuel confidence in using algos and evaluating their performance.
Execution quality and transparency will increase algo adoption in FX. Thanks to the GFXC’s recent changes to the Code, market participants are more likely to trust algo services and rely on TCA to validate effectiveness.
This virtuous cycle, coupled with external regulatory changes such as UMR, will likely increase algo usage in other asset classes. The next few years are set to offer interesting trends in algo usage and adoption.
Bruel is optimistic about algo growth in FX. “Algos will continue to grow where they are used today,” he says, “and also become more prevalent in asset classes that are currently less conducive to algo use such as NDFs.” So the future is bright for algo use and the next few years will undoubtedly bring interesting developments in this space.