Ivy Schmerken

What lies ahead for FX Algo Trading and TCA?

June 2023 in Market Watch

Foreign exchange participants are facing an evolving market structure in 2022 with a renewed focus on algorithmic trading and transaction cost analysis (TCA) tools for institutions executing trades in a complex landscape with fragmented data.

A focus on data and analytics is permeating every decision in foreign exchange trading, according to panelists on the virtual FX Market Structure 2022 event sponsored by Tabb Forum and The Full FX, held on December 1st. From selecting FX algorithms to evaluating liquidity providers, and identifying hidden costs, there is an emphasis on what the data is telling people, said multiple speakers at the virtual event. In addition, recent changes to the FX Global Code of Conduct could provide additional market transparency for the related to algorithmic trading and TCA.

Source: Coalition Greenwich 2021 Market Structure & Trading Technology Study

Here are some of the top trends in algo trading and TCA:

Will FX algos gain further adoption in a post-pandemic climate?

While algo trading increased during the crisis, volumes have subsided since then, so there are questions about whether the trend is sustainable and what are the drivers going forward in 2022.

The pandemic served as a catalyst in convincing buy-side traders to execute FX through algorithms, which the most sophisticated investment firms had traditionally used.

Working from home in an environment they were less accustomed to and lacking resiliency, buy-side traders had to execute orders not only in FX but other asset classes they covered. Although the sweet spot for FX algorithms was executing large orders prior to the crisis, banks saw the algos expand their reach amid rising volume and volatility.

“As the pandemic unfolded, clients seemed to quickly appreciate the fact that algos can be used for smaller sized orders, that they actually are robust, and in times of volatility they are able to give the outcomes that people say they will give,” said a senior bank trading executive, speaking on the algo panel.

When selecting an algo strategy, clients tend to have one of three objectives: either to reduce market impact (save trading costs), minimize market risk, or maximize execution certainty, said the bank executive.

When selecting an algo strategy, clients tend to have one of three objectives: reduce market impact, minimize market risk, or maximize execution certainty

According to a TCA expert on the algo panel, about 41% of FX spot volumes were executed via algorithms at the peak of the pandemic in 2020, and the figure was up to 55% for G-10 currencies over the same period.

Though spreads have tightened, and volumes have decreased, there is more flow executed via algos than pre-crisis, and banks have made headway with adding clients in the corporate space who are the non-traditional users and across all different client segments. Banks with expertise in non-deliverable forwards (NDFs) in emerging markets see that as a growth area for algorithms.

Lower costs and technological advances in connectivity to many bank algos are drivers of algo adoption. On the panel, FXSpotStream, a bank-owned consortium that provides an infrastructure to route client orders to multiple liquidity providers, said it launched an algo offering in July through a single API (application programming interface) and graphical user interface (GUI).

The evolution of the execution management system (EMS) has also facilitated adoption of broker algos and TCA. Many buy-side trading desks use an EMS to centralize the relationship with liquidity providers, connect with algos, access analytics, and capture the fill data and time stamps necessary for TCA.

A recent study by Coalition Greenwich found that 69% of FX respondents believe that algo usage will increase. Though algo adoption in FX is less prevalent than in equities, the research firm noted that market practices change over time. “The FX market has been steadily adopting algos, and we expect that to continue,” wrote Stephen Bruel, Head of the Derivatives and FX Practice on the market structure and technology team at Coalition Greenwich. “Technology advancements in the overall FX market, including new execution strategies and more accurate transaction cost analysis (TCA), could reinforce the trend,” wrote Bruel.

How are the buy side curating liquidity providers and executing with FX algos?

Buy-side trading firms are looking to customize their access to streaming prices from liquidity providers in electronic venues to achieve their best execution goals while minimizing market impact. One execution strategy involving FX algos that has regained the buy side’s interest is known as “full amount trading.” Rather than slicing up an order and trading with multiple LPs, which can lead to market impact, some are moving to execute the full amount with fewer counterparties.

The head of FX sales for a major global bank on a panel noted that conversation used to be about adding more liquidity providers and more streaming to the top-of-book. “Now, as more participants examine the data and understand it, the conversation is about the full FX amount (executed), understanding LPs and which LPs belong in a pool for a certain type of business, and then following up with market impact analysis,” said the FX sales executive.

“Full amount trading is about executing whole parent orders as one placement instead of slicing it up into many placements,” explained Uday Chebrolu, Senior Vice President, Head of Product & Engineering for FlexFX EMS at FlexTrade Systems in response. According to Chebrolu, this is now a standard feature in FX trading and is offered out of the box with FlexFX EMS. Banks also create full amount levels (based on VWAP bands) as desired by the buy side, he said referring to volume weighted average price. Some exchanges and ECNs offer this as well, he noted.

Is TCA reliable and is it evolving to incorporate real-time events?

Panelists at the FX conference agreed that transaction cost analysis tools will be important to higher adoption rates in algorithmic trading. “Historically people are less confident in FX TCA. The next stage is the ability to compare and contrast providers on a like-to-like basis as well as specifically on algos,” said the TCA provider.

One of the hurdles with TCA has been fragmented data, and the challenge of consolidating this data in one place. In an over the counter (OTC) market such as foreign exchange, buy-side traders rely on relationships with different banks for execution and pricing. There is data to collect from multiple liquidity providers and venues. Asset managers that have multiple OMSs and EMSs could receive data in different formats.

Stephen Bruel

“The FX market has been steadily adopting algos, and we expect that to continue,”

Stephen Bruel

Normalizing the data is still a challenge for the buy side, who are often resource constrained and often compete internally for their IT department’s services. “The big challenge with data is to get it standardized so we can do some observations around executions and make sure we are analyzing like- for-like,” said Richard Turner, Senior Trader, Currency Solution, Insight Investment, which uses algorithms to meet the firm’s best execution regulatory requirements.

However, the TCA provider said it’s key for buy-side traders to look beyond their own data, noting that it provides a community data pool. “The starting point is being able look at a broad data set and make decisions on which algo is most appropriate for your flow,” said the TCA expert. Yet, a second provider of execution analytics suggested that using independent data is insufficient and that data from liquidity providers is necessary.

TCA is getting to the stage where it facilitates more informed decisions

How are buy-side firms using TCA?

While many years were spent gathering accurate transaction cost data in FX trading, and ensuring the results were reliable, TCA is getting to the stage where it facilitates more informed decisions, said panelists.

Institutional clients are seeking a deeper understanding of how the algos are going to work. They want to know if an algo traded at a certain time, what is the likelihood of the algo executing to completion, and how much it’s going to cost. If a trade is executed when volatility is high, they want to know the market impact, said the bank trading executive.

“Full amount trading is about executing whole parent orders as one placement instead of slicing it up into many placements,”

Uday Chebrolu

Institutional clients have asked their banks for pre-trade data and real-time analytics. Large banks such, whose teams have tested and run simulations on their algos for their own trading, are providing the in-house data and analytics to clients with their algo suites.

For example, a client might backtest an algo strategy based on 50 days of look-back data to gain insight into how it performed on various metrics they chose, explained the bank trading executive. “Then the trader can run the algo, and it can constantly compare the outcomes against those predicted outcomes from the start as to how it’s performing.”

A trader may learn that trading at a specific time of day did not work for a currency pair, they were too aggressive, or there was not enough liquidity or that volatility picked up as the market changed, said the bank executive.

In terms of flexibility, many tools allow traders to adjust the tools for real-time events, such as news or changes in the market, said panelists. “Traders want to interact in with TCA. Not only do they want to set the parameters and let it run, but they want (TCA) to tell them if there is something to prepare for or what’s the next step of the execution,” said the head of FX sales for a major global bank.

The next step is using post-trade TCA to create a feedback loop into the pre-trade analysis. For example, on the panel, Insight Investment’s Turner said when the firm is executing through an algo and liquidity conditions change, the idea is to feed that data back in. But Turner conveyed that TCA needs to fit into the firm’s automated workflows where trades are untouched and are monitored for exceptions.

The FX Global Code of Conduct will drive algo adoption

“We’re trying to build automated processes where a lot of these workflows become automated to a degree, so we don’t touch them, where we have all these checks and balances to make sure that when we have automated a trade it goes through correctly, and it is within best execution regulations,” said Turner.

“If a red flag goes up, when something is wrong, the trader wants that red flag to go up before the next trade is made.” Turner compared the trader’s role to a pilot at the controls of a plane who can step in when there is a problem. “It’s all the automation and doing all the checks and balances to ensure that our best execution regulation is taken care of,” he said.

When using TCA, traders need to iteratively use the tool to gain better results, advised the third-party TCA expert. “It’s not good looking at the TCA periodically. You need to be putting those ideas into action, and that leads to best execution,” said the TCA provider, who advised users to review results, address outliers, tweak their inputs and then start that process again.” FX participants expect TCA to get smarter and the head of FX sales for a major sell-side bank said it would be embedded in routing and trading decisions.

What is the impact of the revised FX Global Code of Conduct on algorithmic providers and TCA?

After an extensive three-year review, the Global FX Committee (GFXC) amended the code to enhance practices in FX trading in key areas such as last look and pre-hedging, as well as anonymous trading, algorithmic trading, and transaction cost analysis, and settlement risk. It also published templates for use by industry participants to improve disclosures and assist with transaction cost analysis.

Specifically, the revised code asks algorithmic trading providers and aggregation services to publish information on how they operate and to disclose any conflicts of interest in handling client orders. The updated code also encourages algorithmic trading providers to provide a clear description of the algorithmic execution strategy with information to clients on how to evaluate their performance.

In its report, Greenwich Coalition predicted that the FX Global Code of Conduct will drive algo adoption.“While some motivations to use algos remain consistent (e.g., accessing liquidity), “there are new drivers that are changing this part of the market,” states the report summary.

“The big challenge with data is to get it standardized so we can do some observations around executions and make sure we are analyzing like- for-like,”

Richard Turner

A sell-side participant said that large banks support the code, and favour clear, accessible disclosures to promote informed decisions and transparent markets. But the executive insisted that big banks were already operating under a high bar and were making these disclosures to their clients. The panelist suggested the updated code disclosures would help second-and-third-tier providers and promote competition, better technology, and a level playing field. In addition, the revised code recommends that algorithmic trading providers disclose information for use in TCA in a “market-wide standardized format,” and has published the GFXC Transaction Cost Analysis Data Template.

The new disclosures would help to normalize data from a TCA perspective, said the TCA expert. The TCA provider takes in 27 bank feeds and noted that a major gap is ‘reject trades’, which would be an additive to analytics.

Looking ahead, market participants say that FX algo trading is in its infancy, but they predict it will continue to develop as clients become more confident about the outcomes. While TCA is still a work in progress, the new code of conduct guidelines and TCA template could help standardize data inputs across the banks and third-party providers and allow for more apples-to-apples comparisons.