Market Impact in FX: Isolating the problem and treating it

FX’s electronification has helped market stakeholders improve processes while shining a new light on existing problems. Specifically, firms are reviewing their efforts to reduce the impact of their trades, making information leakage and risk reduction hot topics currently.

Ideally, optimal execution will reduce risk while generating almost zero market impact. However, achieving this state is proving challenging to say the least. Oleg Shevelenko, FX Product Manager at Bloomberg, believes market participants often misunderstand what optimal execution means, reducing it to a question of execution spreads.

“Due to its complexity and nuanced nature depending on [the participant’s] and transaction type, it is often misunderstood,” he says. “An asset manager might be trading the tightest price, but those executions could be moving markets.”

While Paul Clarke, Head of FX Venues at LSEG, points out that market impact also affects liquidity. “Trading on a full-amount basis with liquidity providers capable of warehousing your risk will always minimise market impact but in the real world, this is not always possible,” he says.

“For the largest clients, the size /duration of an FX position may be too large for a single counterparty to hold—both credit risk and market risk being contributing factors here.”

So how can firms minimise their market impact, and how are service providers helping them achieve this goal? Examining why achieving a balance between risk and market impact is challenging is a good place to begin.

Why market impact is more pronounced of-late

When asked why the issue of reducing market impact has emerged recently, James Dewdney-Herbert, Associate Director, e-FX at Saxo Bank, says that the topic was always present in the background. “It has always been important,” he says. “Advancements in big data interpretation highlight it more clearly. Market impact diminishes a trader’s alpha and the efficacy of hedging, so its importance cannot be underestimated.”

Bailey White, Director, Electronic Trading and Execution at 26 Degrees, explains that market impact is often tied to a trader’s rate of execution and their interaction with liquidity. “This becomes a complex issue because typically traders who are looking to enter a risk position based on a certain signal, or who are managing market risk can be reluctant to slow down how quickly they enter or exit this risk by adjusting their flow rate” he says. “LPs will widen pricing until they reach an equilibrium between market impact and spread at inception—ultimately leading to the trader paying a premium for their desire to enter or exit risk more quickly.”

Firms often choose LPs with high internalisation rates, combining that with algo executions that space trades over time. While this approach ensures optimal execution, it does little to mitigate market impact.

“By spreading execution over time, the trader will be assuming market risk as the price of the currency can change,” Shevelenko says. “Effective balance of managing market risk and market impact as a total cost of trade is the art of achieving best execution.”

Dewdney-Herbert echoes this point and says, “When we work with high volume, high frequency strategies we aim to mitigate the market impact by transacting with liquidity providers who internalise rather than transfer the risk. The latter magnifies impact which is why serious liquidity users are choosy about who they transact with.”

LSEG’s Clarke says that minimising market risk and reducing market impact are not necessarily opposed to each other. “If you’re trading with algos, absolutely your choice is duration, which is market risk, versus impact, which is speed,” he says. “I can trade on a full amount basis, I can trade with other counterparties, I can use different frameworks to trade which don’t create these offsets in quite the same way.”

Clarke points out that the best algo providers manage the balance between market impact and execution speed by understanding the liquidity environment. Mismanagement in this regard creates an unwanted market impact. “I think electronification has helped but also hindered,” he says. “The biggest innovations in platforms recently are ones that promote liquidity aggregation for large orders without managing the pool of providers you execute with.” 

While these tools are useful, they give stakeholders a false sense of confidence. “If I have a liquidity-seeking algo attempting to buy too large an amount, I’m going to have an enormous market impact,” Clarke says. “If I try and trade a TWAP for an amount over too short a period, I’m going to have an enormous market impact.”

He says that these tools have effectively moved the problem of “who am I trading with?” to “how am I trading and being sensitive to the environment?”

John Marchese, VP, Head of FX EMS Sales, Americas, FactSet, points to another factor when highlighting how electronification is increasing market impact. “Market-making technology has advanced quite dramatically,” he says. “Banks that have budgeted for faster pricing and trading technology could potentially benefit from the winner’s curse scenario that plays out on larger RFQ/RFS trades.”

He explains that if a bank loses a bid in a competitive RFQ/RFS, its system knows that risk will likely enter the market very soon, presenting an opportunity to front-run that risk.

“By and large, most (if not all traders) worry about this [reducing market impact] daily,” he says. “They are always looking for more insightful analytics to better understand if their executions are signalling the market and potentially reducing the performance of their next trade execution.”

Dewdney-Herbert is forthright with his views. “Electronification allows FXLP’s to operate like agency brokers on steroids,” he says. “Rapid risk transfer, aiming to capture decipips between client fill and street hedge (not necessarily in that order) increases impact as the hot potato changes hands before meeting real interest.” 

26 Degrees’ White terms this “fragmentation” and says this is a consequence of electronification. “Improvements in technology mean the LPs have data to show you where a trader or broker has directly caused market impact via their activities,” he says. “As such, pricing of different flow profiles has become more disparate as LPs increasingly segregate impactful flows from benign flows.”

Continuing this theme, Bloomberg’s Shevelenko says, “Market participants split their large orders and execute across multiple venues. Advances in technology allowed those trading venues to decrease market data publication times, down to single-digit milliseconds, creating an environment where trading patterns can be easily detected.” 

Measuring market impact

While every trader understands the market impact consequences of their decisions, opacity still surrounds the subject. “The FX market is still early in its journey to create a common cost model to reliably calculate and attribute transaction costs to factors and evaluate their consequences on market impact,” Shevelenko says.

He points to liquidity fragmentation across venues and dealers as a stumbling block. This fragmentation results in a lack of statistically valid datasets that can feed mathematical models.

Another challenge looms for the FX market, aside from fragmented data. US Equities will move to T+1 settlement, creating interesting questions from a market impact perspective.

Clarke believes that while liquidity close to the date rollover will be challenging to source, the presence of high demand (should it materialise) will ease these concerns in the medium term. 

“A consequence of a lot of trading happening there means that there’s a lot more interaction between market participants, which means that liquidity begets liquidity,” he says.

One consequence is that firms will look for better ways to measure the market impact of their orders. Factset’s Marchese explains that measuring market decay post-trade is one way of quantifying market impact. “Our clients see a full view of their risk transfer price via Executable Streaming Price (ESP) connectivity which is always running in the background,” he says. “The buy-side has the option of trading on the ESPs or using that data to help analyse any market decay resulting from their completed algo trade or from a standard RFQ/RFS competitive execution.”

He explains that the buy-side also uses arrival price comparisons as a means of measuring both execution costs as well as market impact implications. “Over the years we’ve found that buy-sides identify “arrival price” slightly differently, either based on order arrival to their execution trading blotter or the price of the currency pair in the market when they initiate an execution process,” he says. “Both can be measured effectively and provide total cost of trade and impact metrics.”

When asked what is the best way for firms to measure their market impact, Clarke says, “I think the only way you measure market impact is pre and post-trade markout analysis. A minute before you issued an RFQ, what was the market doing? Was it neutral? Was it moving up? Was it moving down?”

These observations extend to after issuing an RFQ. “Has the underlying trade trend of the market continued while that RFQ is out and you’re getting prices at the point you execute now, what happens?,” Clarke says. “Does the market then jump?”

Given that this analysis happens on a per-trade basis, gathering data and observing impact over time is the best way forward. However, isn’t this a cumbersome process for the average firm?

“The easiest way firms go about doing this pragmatically is to use a third-party provider that specialises in this,” Clarke says. “There are several TCA providers that do it. Use peer analytics to [analyse] whether you are doing better or worse than others.”

Saxo’s Dewdney-Herbert says, “Market impact can be tricky to measure because it’s relative to liquidity consumption. The solution from a buy-side perspective is comparing implementation cost front-test outcomes versus actual TCA. If you’re overshooting your implementation cost target and your other cost components are all in line, then you’re experiencing too much impact.”

White notes that while some third-party vendors help firms measure the market impact of their trades, this data is useless if not combined with effective liquidity management. “Prime of Primes can assist in this regard,” he says.

“26 Degrees has developed a proprietary data analytics system (“Insights”) in-house that measures market impact across all client activity real-time. The data can be readily grouped and filtered across multiple facets of client activity (time, pair, size, client-tag, order type, market condition, correlations etc) which allows our team to communicate openly with clients to easily identify subsets of higher impact order flows, changes to market impact profiles, and trends across liquidity pools/providers.”

When asked how trading platforms can help firms measure market impact, Shevelenko says that bringing the issue to the forefront is critical. “Trading platforms need to provide their buy-side and sell-side participants tools to measure market impact over time,” he says. “Highlighting the problem and bringing it to the forefront is a big step towards addressing appropriate behaviours.”

White adds, “Platforms have continued to enhance their in-built analytics suites and the ease at which clients can access their own ‘raw’ data. This lowers the barrier to entry to managing market impact. It more readily allows takers to understand the value of their flow & the consequences of executing in a manner that does induce market impact.”

Benchmarks go a long way toward helping firms understand their degree of market impact. To this end, Shevelenko mentions that FXGO developed a series of best execution benchmarks for active traders. “The types of benchmarks include cost of trade, cost of rejects, market impact, reject ratios, and dealer responsiveness,” he says. “As the FX market is relationship-based, such common metrics allow clients and dealers to have a discussion around optimising trading behaviour and pricing to improve best execution.”

Clarke notes that aside from offering benchmarks, “What any solution needs to ensure it can solve for is the needs of the buy side participant in executing at a benchmark, and liquidity providers in being able to fill their clients at that rate effectively.”

He also explains that firms must examine whether a benchmark makes sense. “If you receive an order at 10:00 a.m. London time and your benchmark is to trade at the four PM WM fix, that could be inappropriate,” he says. “There’s a lot of trading there that could happen. Is it appropriate to use that?”

NEW LIQUIDITY POOLS

Factset’s Marchese says that new liquidity pools are entering the market with the sole purpose of reducing market impact. “These new liquidity destinations must have the total support of the sell-side community or else they will never realise their full potential,” he says. “New entrants like LoopFX aim to solve this concern by allowing the buy-side to discreetly check their relationship liquidity to match at a mid-price without the market knowing the buy-side’s initial intent.”

Shevelenko notes that combining algo execution with these selected liquidity pools goes a long way toward reducing market impact. However, he cautions against thinking that algos are an automatic solution to optimal execution, echoing Clarke’ comments. “Poorly designed algos or any automated execution methods with easily detectable trading patterns can themselves create a “signalling risk”, underlying the point that trading tools need to be carefully selected and evaluated,” he says.

Clarke says that algo execution offers consistency by removing subjective inputs from an execution decision. That alone removes elements that can create market impact. “But the process matters more than the tech,” he says. “It needs coupling with pre and post-trade analytics and decision support. And once I’ve made a decision, what do I do? Which strategy do I use? Which set of parameters should I use? So it [algo execution] does help, but applying a scientific method to the overall process is important.”

White notes that progress in this area has been significant. “The current suite of market-leading algorithms all focus on limiting market impact and information leakage,” he says. “These algorithms now also incorporate mid prediction logic to reduce adverse selection costs and enable an increased execution speed.” 

Of course, the algos aren’t a solution by themselves. “Managing this in-house efficiently requires an experienced and dedicated team, and access to an appropriate pool of liquidity providers who do not cause information leakage,” White notes.

Meanwhile Marchese believes stakeholders can use algos in another way. “Broker-neutral proprietary algos can drip-feed larger deals into the market, so a client’s position isn’t being broadcast, which can minimise information leakage.” he says. Additionally, he adds that the buy-side can set up a peer-to-peer/dark pool algo wheel that functions as their first stop for sourcing liquidity. “They could potentially get some, most, or all, of their trades done at a mid-rate. Any residual amounts could be subject to price negotiation with a single counterpart (to keep impact low) or the buy-side desk could take the balance of their order to the street in a competitive RFQ/RFS.”

Technology’s role in further reducing market impact

Technology, in the form of algo execution, gives stakeholders a framework to reduce subjective decision-making as Clarke says. But there’s more to technology than just algos.

Bloomberg’s Shevelenko says that tools like trade cost models can help traders measure anticipated market impact as a component of the overall trade cost for a given order size and execution duration. “Subsequently, such models will allow traders to evaluate their deviation of the expected impact and adjust their trading styles and liquidity selection accordingly,” he says. “Wide adoption of such tools would lead to self-correcting orderly markets.”

Marchese offers an example solution Factset’s clients use currently. “Our clients have a toolset at their disposal to reduce trade size through the slicing of orders (manually or automatically),” he says. “In more liquid currency pairs, this could shade the larger order and reduce any signalling to the market and thus reduce the impact of the order.”

Interestingly, he says that leaning on FX’s relationship-based structure is still the best way to reduce market impact. But isn’t this a decidedly legacy-like approach? “Even in today’s market, “old tech/low tech” is the main means to reduce market impact,” he responds. “Good relationships, phone orders, and chat sessions still rule the large notional trade market. Although algos have held a role here, FX still is heavily relationship-based.”

Clarke believes that the information needed to reduce market impact is already there. The question is, how will firms adopt it, especially ones that do not execute frequently? “Imagine a corporate that executes once every year as part of a very large cash flow hedging exercise,” he says. “They may not apply the same approach as the others and it’s probably to their disadvantage. So the question is, how do we make sure that all participants benefit from this information and not just this subset that is sophisticated, and do this regularly? It’s a case of promoting adoption rather than building more widgets.” 

Shevelenko says using technology to produce metrics that allow firms to change their execution methods is key. “By producing common metrics to indicate potential signalling risk for different execution methods, liquidity pools and providers can allow market participants to make informed trading decisions,” he says. 

“As the time to make trading decisions is quite short due to increased trading volumes, it is also key to make those tools available at the trader’s fingertips via execution management systems and trading platforms.”

“Information leakage is by no means a challenge that has been entirely resolved in the market.” White concedes. “Detecting information leakage is still a challenging process and we find that transparent communication plays an important role in identifying clients and venues that are not skew-safe.”  

Ultimately, technology’s role lies in offering greater assistance during the times when firms need it. While it may not offer a golden solution to reducing or eliminating market impact, it is helping stakeholders design efficient processes that achieve this goal.